Home

- I-Cube

Contents

1. jl F i T Iam i LE Tim me LTE F Ee Ta Sam LEE f I m fon E F e Fh T las a EE H Som i m a M e Ta log Fo j LG k fie i on a 2 OE a g i pa ka Fo l ber tb m foe E Eg DUDLEY BARRY 250 imagets D image s selected a i aa A 3 1 t 4 ai itis i SR 11 ad HH AIH WH AH lie ENIH H Confidential Power Automation Printed 5 29 2005 WEE aihicden ace fey BE 21 FACIAL Page 59 of 104 Dawa Sewbalak IDENTIFICATION SOFTWARE USER MANUAL 21 1 How to use I CUBE Facial Search 21 1 1 Enroll You can enroll face from static image file or live video To enroll from static image file click Get Photo From File button and select the image file We currently support JPEG BMP WMF and TIFF formats To enroll from vide you have to enable video if there is no video displayed You can do it by clicking Video button which is located on left panel Make sure you have camera and driver installed before doing that You can adjust the video format and video source by clicking Video button again a dialog box Video Settings will appear you can also turn off the video in that dialog box After video is on click Start Track to start tracking Click Get Photo From Video after face was tracked Enter personal information You have to enter at least First Name and Last Name Click Enrol button It will pop up an enrollment successfully message box if everything
2. Once installed the system will be commissioned All training will then be finalized to ensure that staff are fully appraised of the operational and support issues The system will then be made live and operationalized Final project sign off will then take place Confidential Power Automation Printed 5 29 2005 JT ETF AFETE E Ap TPA k Se HCUBE 9 EQUIPMENT Considerations Page 22 of 104 Dawa Sewbalak System description Full featured vehicle access control system with license plate reading software linked to facial identification and verification Item PC P4 3G Hz 512MB RAM Windows XP 19 SVGA Monitor Keyboard and Mouse Switch Cabling per metre supply and install Commissioning Software to integrate LPR to FRS SUB TOTAL NPRS ART s LPR identify vehicles by a combination of color shape texture and license plate I Cube Stand Alone System I Cube SEE ROAD LPR Software 1 site 1 lane mono Inc Dongle Lic I Cube SEE ROAD LPR Software 1 site 1 lane STEREO Inc Dongle Lic I Cube SEE TRAFFIC narrow lane 1 See Traffic Head lane I Cube SEE TRAFFIC wide lane 1 See Traffic Head lane I Cube DLL LPR Software 1 site unlimited lanes Inc Dongle Lic GEOVISION LPR Software 2 lanes 5 cameras Frame Grabber Card I O Cards 7 lanes x 2 inputs 14 4 input LPR cameras REG LED 7M LPR Camera I CUBE Camera IR LPR Camera ZC NH 403NP Day Night 480 TVL Camera 12v 2
3. Enter the name of the file to which the activity log 1s to be saved If an existing file is selected it will be overwritten Click the OK pushbutton Activity log files that are saved in text ASCII format are assigned a log extension Power Automation Printed 5 29 2005 S OE aT GT ne CUBE Confidential Page 54 of 104 Dawa Sewbalak The list of events provided within a window of the Activity Log Viewer can also be saved to a binary file Activity logs saved using this method store the associated images in JPEG format Binary activity log files may also be reloaded back into the Activity Log Viewer unlike the Activity Logs saved using the text file format To save an activity log to a binary file perform the following operations Select the window containing the activity log that is to be saved Select the Save Binary File menu item from the File main menu A standard File save dialog will appear Enter the name of the file to which the activity log is to be saved If an existing file is selected it will be overwritten Click the OK pushbutton Activity log files that are saved in binary format are assigned a bin extension Note that a portion of an activity log may be saved in either text or binary format by multi selecting the activity events records displayed on the left panel and using the right click pop up menu Images that are displayed on the right panel of the selected window may be saved to JPEG file
4. NFA is the number of false acceptances NHA is the number of impostor identification attempts NIVA is the number of impostor verification attempts False Rejection When a biometric system fails to identify an enrolee or fails to verify the legitimate claimed identity of an enrolee False Rejection Rate FRR The probability that a biometric system will fail to identify an enrolee or verify the legitimate claimed identity of an enrolee The False Rejection Rate may be estimated as follows FRR NFR NEIA or FRR NFR NEVA where FRR is the false rejection rate NFR is the number of false rejections NEIA is the number of enrolee identification attempts NEVA is the number of enrolee verification attempts This estimate assumes that the enrolee identification verification attempts are representative of those for the whole population of end users The False Rejection Rate normally excludes Failure to Acquire errors Field Test Field Trial A trial of a biometric application in real world as opposed to laboratory conditions Filtering The process of classifying biometric data according to information that is unrelated to the biometric data itself This may involve filtering by sex age hair colour or other distinguishing factors and including this information in the database Goats Biometric system end users whose pattern of activity when interfacing with the system varies beyond the specified range allowed by the system
5. eliminating inefficiencies and unproductive resources We also provide customers an efficient way to secure their business environment reduce costs and ultimately drive profit Information Technology has become so integral to success that it is now not only a support function but could plays a proactive and vital role in realising the business role We offer technology solutions and services that allow customers to efficiently integrate manage and maintain their people processes and assets Our adaptive infrastructure becomes increasingly critical to business success balancing agility robustness and affordability By sidan the problem we can address the r VIR solution By reviewing what is occurring around the world and customising this to your specific needs we provide a complete solution The cost of technical solutions to maintain the gap ahead of the competitors and to continue fighting crime is expected to keep rising steadily over the next few Confidential Power Automation Printed 5 29 2005 Ue Ud HCUBE Page 28 of 104 Dawa Sewbalak years To compete for investment dollars we allow you to employ techniques like infrastructure pattern matching reusable design infrastructure impact assessment analysing reuse predictive cost modelling total cost of ownership and budgeting and application subscriptions service level oased packaging Moreover users cannot afford to wait until full deployment for a q
6. file name with full path there You can still update the field ordinates at this time if you find you entered wrong field ordinates Just press Update button after you updated Press Start to start batch enrollment if you feel everything is all right By default it will start enroll from 1 record You can change that by typing number of records to start and click GoTo button Confidential ebay Dae a a fi Hai ia Er Ten rs Power Automation Printed 5 29 2005 eT NED TT ee CUBE before starting batch enrollment After batch enrollment is finished it will give you statistics on number of images enrolled successfully Some of images you may need mark eyes manually You can do this by clicking Mark Eyes button It will go through the database to find out those unmarked images Note those unmarked images are not searchable before you mark them manually 21 1 3 Search You can search the image from image file video or database To search image from image file click Get Photo From File button and select the image file To search image from video enable video click Start Track button and click Get Photo From Video after face was tracked Page 61 of 104 Dawa Sewbalak To search image from database click Get Photo From Database button and select the image in database You can search more that one image to ge
7. in order to generate the facial biometric template The greater the volume and quality of the enrollment images the faster and more reliably the system will recognize the user during subsequent verify or classify operations Enrollment is performed by clicking the Enroll button on the control bar of the POWER AUTOMATION Facial VERIFICATION main window User Registration Within the POWER AUTOMATION Facial VERIFICATION system a user may be registered before they are enrolled This means that users may be entered into the POWER AUTOMATION Facial VERIFICATION database without enrollment of facial images or storage of an associated biometric template Registration may be performed for one user at a time through a dialog or for many users using an ASCII text file Registered users are automatically enrolled the first time they present their ID token i e proximity card keypad through a Wiegand device or enter their user ID manually through the Enroll control button Template Generation Biometric templates are generated and continuously updated through a process referred to as Training using the facial images captured during the enrollment operation Further enrollment i e capture of additional facial images may be performed during subsequent verify operations This ensures that the biometric templates are as up to date as possible Confidential Power Automation Printed 5 29 2005 viriai ances GUBE Page 21 of 104 8 3 Normal p
8. incorporated into your system POWER AUTOMATION specialises in integrated digital CCTV Management systems that are designed according to each client s unique requirements and specifications Intelligent digital CCTV management systems by POWER AUTOMATION _ presents operators only with the information that they need to re act upon and provides management with instant retrieval of vital recorded footage The POWER AUTOMATION option is not simply a security system it allows management to focus on productivity improvement loss control safety management procedure audits equipment maintenance time and motion studies and ensures that managers and security staff spend time where they are needed most in their operational functions In the rapidly changing digital world it is important to understand that information and the rapid and efficient interpretation Type of LP System amp Figure 23 USE OF LOSS PREVENTION SYSTEMS amp PERSONNEL Personnel Burglar Alarms Live Visible COTY Check Approval Database Somening Systane Armored Gar Deposit Pickups abies Looks Chains Live Hidden OOTY Digital Video Rroording System Observation Wiros POS Data Mining Sutea Secured Cisplay Fixtures Crop Sates Walen Hones Shoppers Acousko Wagneto Electronic Seourity Tage Shapifing Deieren Signage Uniformed Gus rhDye Benefit Denial Tags ike Alama Simulaled Visible COTY RF Bedonia Seriy Tags Aai Gothes Store Det
9. input 110 220 VAC YAW SCREW PITCH SCREW Figure Pitch amp yaw screws Confidential Power Automation Printed 5 29 2005 Dawa Sewbalak Page 38 of 104 The standard Infra Red model specifications are Camera Sensor Scan Resolution Shutter Power Lens Illumination Spectrum Angle Intensity Power Effective Range Physical Case Standard Temperature Degrees of freedom Attachment Dimensions Drawings Electrical Power Inputs Output CCD 1 3 B amp W CCIR 625 Line interlaced 380 TV lines 1 1000 9 16 VDC 100mA F1 2 16C 8 12 16 mm a near Infra Red for most Countries 30 3 levels pulsed 12VDC 3A pulsed 8M reflective plates to 4 5M non reflective Enforced Poly Carbonate UV protected IP 65 weatherproof 10 c to 50 c 2 left right up down 2 x 8 mm screw Front 150x150mm Depth 135mm 35mm hood Arm 160mm can be downloaded from support page Power Supply 3A 15VDC 2 lines TTL 3 levels of intensity off Composite Video 1Vp p 75 Figure Front amp Rear installation dual cameras per lane shared activation e Input Output control card including flat cable and terminal block with cover TO INSTALL three 3 PC 19 SVGA MONITOR Confidential Power Automation Printed 5 29 2005 Ue Ul HCUBE Page 39 of 104 Dawa Sewbalak e AN UNINTERRUPTED POWER SUPPLY FOR THE SERVER e OPTIONAL EXTRA ELECTRONIC DISPLAY
10. 1 x MS 24 6 2L 1000 mcd e This unit consists of 2 lines With 24 characters per line Colour RED LED s IP Enclosure The PORT ACCESS CONTROL welcomes INTEGRATION OF ALL ITEMS DETAILED Confidential Power Automation Printed 5 29 2005 PL bo a ee ee ns kH a HCUBE Page 40 of 104 Dawa Sewbalak 15 Software Quote The software is compatible with Microsoft Windows 2000 Professional and Microsoft Windows XP Professional The minimum system configuration requires a video capture card compatible with DirectX 8 0 in addition to the standard PC hardware Minimum hardware requirements are listed below I CUBE Facial Identification Client Microsoft Windows 2000 Professional Service Pack 4 or Microsoft Windows XP Professional 1 GHz Pentium 4 Processor 128 MB RAM 10 GB HDD CD ROM Drive WDM compatible video capture device I CUBE Facial Identification Server Microsoft Windows 2000 Professional Service Pack 4 or Microsoft Windows XP Professional 2 6 GHz Pentium 4 Processor 512 MB RAM 160 GB HDD CD ROM Drive WDM compatible video capture device The performance of the I CUBE Facial Identification CLIENT PC is subject to at least an ISDN connection to the central server The I CUBE Facial Identification requires a skilled operator who has either been trained or has read and understood the I CUBE Facial Identification user manual The software will allow the following info
11. 2000 XP All specifications are subject to change without notice All brand names and trademarks are the property of their respective owners 8 2 Additional Items All events will be captured on a digital video recorder to allow access to real time and historical records Confidential Power Automation Printed 5 29 2005 ee ee ns k a i Page 20 of 104 Dawa Sewbalak 16 Explanation of Biometrics In order to work with the POWER AUTOMATION Facial VERIFICATION system it is beneficial to understand the basic concepts of system operation Each User that is registered within the POWER AUTOMATION Facial VERIFICATION system has an associated facial biometric template which contains the information based on enrollment images used to identify the User Biometric access control relies on three mechanisms enrollment of the users biometric data facial images generation of the biometric templates using the enrolled facial images and subsequent VERIFICATION of the user applying the biometric template Tracking In order to make face VERIFICATION non intrusive and flexible the POWER AUTOMATION Facial VERIFICATION system automatically locates and follows any human face that is within the camera s field of view This allows the individual to act in a natural manner with freedom of movement and locomotion and minimal cooperation with the system Enrolment Enrollment is the capturing and storing of facial images of the user
12. Automation Printed 5 29 2005 MiP REN Es aT aT se en CUBE Page 30 of 104 Dawa Sewbalak Throughput rate requirements for both enrolment and operation will affect the successful implementation Almost all systems require enrolment with some eennignes requiring multiple enrolments One will have to provide Tan Lat tL EAA Puen Fe a personnel for the use of the exclusion pe ee ND ote Pw technique during operation to observe or operate the system and users 12 4 Operations The application of security and monitoring sb techniques as proposed will lead to an increase of core competence The security department needs to grow with the use of surveillance equipment this is a subsystem and the combination of skills processes technologies and assets which come together within each subsystem to confer sustainable repeatable and unique competitive advantage It is essential to plan and execute new categories which continue to build and reinforce these competences The solution proposed allows this to occur growing as the security department gains confidence in the equipment and the application thereof The security department has to remain current with the external threats posed such as tips on how to avoid detection by the surveillance cameras Tamburin 2003 Please see the MBA dissertation Casino Exclusion Technique Exploration Framework Development by Barry T Dudley for more information Cameras used to potentially recogn
13. CUBE Confidential Page 52 of 104 Dawa Sewbalak It is possible that fewer than the number images as specified within the Images textbox may be collected during the enrollment operation This will occur in the event that the enrollment timeout limit is reached These timeout values may be modified through the Enrollment tab of the Options dialog refer to Section 3 2 3 When the image collection operation has completed the following dialog is activated to provide manual review and deletion of the enrolled images The Review Enrolled Images dialog provides the option to remove facial images either of poor quality or corresponding to the wrong individual The user may scroll through the collected images using the lt and gt buttons Images are removed one at a time by selecting the image and clicking the Remove pushbutton When the user is satisfied with the set of enrolment images click the OK pushbutton to transfer the images to the FRS database and generate or re generate the associated biometric template The system will collect facial images more quickly if the individual that is being enrolled physically moves preferably by walking towards the camera It is also preferable that the enrolled individual provide a range of head orientation and expressions that is typical of casual movement Physical movement of the individual is useful in that these range of conditions occur naturally and that lighting conditions can be adjusted
14. Database screen the operator may remove any subset or all training images for the selected user The set of registered users is shown in the list box located on left side of the screen Users are listed by lastname firstname and ID number The name of the currently selected user is shown in the Select User box The facial images that are stored for the currently selected user are shown in the right panel The list of enrolled users may be sorted by clicking on the Name or Card column header Clicking on the Name column header will sort the users alphabetically by name in ascending or descending order Clicking on the Card column header will sort the list by Card number in ascending or descending order Confidential Power Automation Printed 5 29 2005 BTF REE Ue es ee CUBE Page 82 of 104 Dawa Sewbalak To select a different user for display and editing click the associated name in the list box or enter a name in the Select User box and press ENTER Users may be searched by Name or Card number To search for a specific user by Name click on the Name column header to pre sort the list then enter the users last name or the beginning letters of the last name in the Select User box and press ENTER To search for a specific user by Card click on the Card column header to pre sort the list then enter the user s card number or the beginning digits of the card number in the Select User box and press ENTER To edit the users info
15. Image Parameters determine the upper limits applied to manipulation of training images 22 2 7 7 1 Offset The maximum amount of offset applied to facial images over both the horizontal and vertical axis This is specified as a percentage of the image size 22 2 2 Resize The maximum degree of resizing applied to facial images This is specified as a percentage of the original image size 22 2 7 3 Rotation The maximum amount of rotation applied to facial images specified in degrees 22 3Server Database The database of facial images may be viewed and edited only from the Discovery Server machine The operational features provided under the database menu item are e Viewing and deletion of images stored within the FRS database and applied during biometric template generation e Clearing and regeneration of all biometric templates stored within the database _ e Batch pre registration of users from an Thr ASCII text file e Enrollment of users from static J PEG ZRERER ERE RG images Gee RR EREEE 7 EZR RR ERE Eo 22 3 1 Editing the Database HPERERERERE ERED The Edit Database utility allows the operator to view and a EENG gg delete images stored within the Discovery Server database a B a a z a 2 naa and used in generation of biometric templates To enable ZRRR RRR this screen select Edit Database from the Database SERRE ERR AS menu MER RREREREEE Figure 22 Edit Database window From the Edit
16. NOT be added to user specific speech strings If deleting a speech string delete all text within the associated record of the database table entitled Hello Table and delete the Wav2 entry for the same record 22 8 Video Settings Select Video Settings from the View menu to change video capture settings Figure 31 The Video Settings window listing available devices The video capture drivers provided by the I CUBE FRS Discovery installation disk are shown in the above list box The appropriate video capture driver is selected by highlighting the driver label and clicking OK Pushbuttons Source and Format bring up the driver settings forms that are specific to the selected driver These forms vary dependent upon the video capture hardware that is installed Refer to the video driver support documentation Select Input is used by certain drivers for selecting either S video or Composite input FRS Central Server The FRS Central Server provides the basic face recognition functionality to the Discovery system It must be running in order for the Discovery system to function 22 9FRS Server Manager The FRS Central Server is installed and run on a single computer While the FRS Central Server is running the Server Manager icon appears in the application tray at the bottom of the screen To change the FRS Central Server settings or shutdown the Server double click on the FRS Server Manager icon The following FRS Serv
17. Register Users window will appear Figure 25 Pre register Users dialog Confidential Power Automation Printed 5 29 2005 pha giny pg C pair i L s ETF AFETE E aT Loar Ibil Sane ECUBE Page 84 of 104 Dawa Sewbalak For pre registering a single User e Enter the User s first name in the First Name box e Enter the Users last name in the Last Name box e Enter the last 5 digits of the User s card number in the Card ID box e Click Insert The single User will be added to the database however a biotemplate or enrollment images will not exist for that User To pre register a group of Users e Click the Select File button An Open File dialog control will be shown e Select the name of the file containing the User information and click Open e Click the Pre Register button to import the list of User names and ID numbers The file containing the Users must contain 7 bit ASCII text only Each line must be terminated by a carriage return line feed pair Each line in the text file represents one User Each line contains last name first name and card number delimited by commas as follows Sample User file Chen Jack 38169 Poss Philip 38163 Sandler Gregory 22283 22 4 Enrolling from Static Images The Discovery system includes facilities to import static JPEG images from a file and apply these images for enrollment template generation Ideally multiple images should be enrolled in order for the system to p
18. Restart button The Central Server will initiate operations and the status display will change to Started 22 9 3 Viewing Connected Client Computers To see a list of all Client Computers currently connected to the a FRS Central Server click the ellipsis button on the Client MESTE D Computers line The Client Details window will appear The Client Details window shows a list of all Client Computers currently connected to the FRS Central Server Each item in the list shows the Client Computer name and the time when it was connected 22 9 4 Managing FRS Training Servers Training servers are used in the generation of biometric templates from enrolled facial images Click on the ellipsis button Contrary Tran on the FRS Training Servers line The Training Server Stg Details window will appear In the Training Server Details window you can view the FRS Training Servers currently connected to the FRS Central Server add and remove Training Servers and control Continuous Training on each Training Server Having more than one Training Server connected to the FRS Central Server decreases the amount of time it takes to train generate templates for all individuals stored within the FRS database The FRS Central Server automatically balances the training load between Training Servers 22 9 4 1 Adding a Training Server To add a Training Server to the FRS Central Server enter the name of the Training Server in the
19. Screen View add Owe By switching to Full Screen view the video window will expand to fill the ee eee entire display area All menus and controls will be hidden Ext In Full Screen view the user has control over a subset of the Standard view s operational features These operational features include ack Ode xi e Classify Mode a ia nt Le en Cans ETEF e Verify Mode Genel Ticking ImsgeFiter Ereinen e Add New User Continuna Training e Show Closest Matches e Show Options en Lee e Exiting the Application eghiness Cordy Database editing and display features are not available M Enabled in the Full Screen view Hyaa DZ To exercise the features in Full Screen view right click m anywhere on the screen A popup menu will appear T Enab Spiich Select the desired feature from the popup menu Figure 8 Popup menu for full screen mode To return to Full Screen view press the ESCAPE key While in Full Screen view Closest Matches mode will display the closest matches on the right side of the screen Confidential Power Automation Change Seve Hane Enter Pensraond Canin Paco Fieiue Deisi Printed 5 29 2005 Ls Cm poir a Te T Ap ee CUBE Page 69 of 104 Dawa Sewbalak 22 2 Setting Operating Parameters The options form allows the User to adjust various operating parameters of the Discovery system This form is enabled by clicking the Options button outlined in the
20. amp EXIT e Where NOT PRESENT in the database access MONITORING of driver face and license plate will occur on ENTRY amp EXIT e Allowing matching of license plate with original drivers face on exit e Each gate will be able to handle a vehicle every 3 seconds per lane e Each gate will be able to manage over 750 000 users in the database plus an unlimited number of vehicles which are not in the database e The GUI will report if o The vehicle is not registered o Inthe database but not allowed access o Inthe BLACK LIST o Vehicle and driver match allowing boom to open e The GUI will allow each vehicle to be granted access based on gate number driver and various other criteria as required please specify e Reports are detailed and comprehensive including o Counts Vehicle registration no s Gate history Date history Vehicles in the database visitors and other categories o Watch list o Ete e The 2 means of identification can be colour shape size card or fingerprint e Integration to parking ticket systems can be provided e All technical specifications are exceeded and detailed below O O O The requirement is to provide Licence Plate Recognition LPR technology NUMBER PLATE vem dealin ei Harpe File linked with FACIAL VERIFICATION to facilitate ACCESS CONTROL of allowed vehicles and associated drivers entering and exiting the EDUCATIONAL FACILITY Confidential Power Automation Print
21. and who consequently may be falsely rejected by the system Identification Identify The one to many process of comparing a submitted biometric sample against all of the biometric reference templates on file to determine whether it matches any of the templates and if so the identity of the enrolee whose template was matched The biometric system using the one to many approach is seeking to find an identity amongst a database rather than verify a claimed identity Contrast with Verification Impostor A person who submits a biometric sample in either an intentional or inadvertent attempt to pass him herself off as another person who is an enrolee Live Capture The process of capturing a biometric sample by an interaction between an end user and a biometric system Match Matching The process of comparing a biometric sample against a previously stored template and scoring the level of similarity An accept or reject decision is then based upon whether this score exceeds the given threshold Multiple Biometric A biometric system that includes more than one biometric system or biometric technology Confidential Power Automation Printed 5 29 2005 ee ee ns Page 99 of 104 Dawa Sewbalak Neural Net Neural Network One particular type of algorithm An artificial neural network uses artificial intelligence to learn by past experience and compute whether a biometric sample and template are a match Performance Criteria Pre determi
22. cues are based upon image variance movement and color The options dialog for the tracking parameters are shown below 5 tecc Figure 11 Options window Tracking tab A description of the various parameters that may be adjusted for the facial tracking operation are described as follows Confidential Power Automation Printed 5 29 2005 C3 co J Ls Cm poir a TP T Ap ee CUBE Page 72 of 104 Dawa Sewbalak 22 2 3 1 Parameters Panel The facilities provided within the Parameters panel allow the operator to set tracking sensitivity and the range of facial sizes for tracking operations 22 2 3 1 1 Follow Threshold Determines the confidence threshold associated with the knowbot structure for determining whether the segmented region corresponds to a face If the confidence remains above this threshold the knowbot will continue to follow the face on subsequent video frames 22 2 3 1 2 Maximum Size This text box sets the upper size limit for faces that are tracked This size limit is applied for tracking during all modes of operation i e enrollment classify and verify The size of the facial tracking box is established as a fraction of the height of the video frame 22 2 3 1 3 Minimum Size This text box sets the lower size limit for faces that are tracked This size limit is applied for tracking during all modes of operation i e enrollment classify and verify The size of the facial tracking box is est
23. disabled by again clicking on the Classify button When Classify mode is enabled facial tracking is initiated automatically After disabling Classify mode the system will remain in Track mode Click on the Track button to turn off facial tracking 22 1 2 3 Show Closest Matches Click on the 5 Close button or select Closest Matches from the Control menu to display the 5 closest facial matches from the list of users currently enrolled within the system The system must currently be operating in Classify mode to enable the display of closest facial matches The closest matches are displayed on the right side of the screen and are shown in descending order of recognition confidence A portrait image of each user within the closest matches list is displayed along with their names and the associated recognition confidence values Click on the 5 Close button again to disable display of closest matches The system will remain in Classify mode 22 1 2 4 Verify Mode Click on the Verify button or select Verify from the Control menu to enable the verify iD Number 22222 operation one to one identification cancel Confidential Power Automation Page 66 of 104 Dawa Sewbalak After clicking the Verify button the Verify window will appear Enter the users ID number in the Verify window and click OK The verification operation will then commence The user will be tracked and compared against the corresponding user that has been p
24. fully at card reader if present Takes a full frontal facial image for 40 msec fora facial verification and another image single image of the VEHICLE capture If the transaction is allowed system If face or license plate or card or car proceeds to open boom colour does not match the information on the ticket control room is alerted and presented with facial image and general car overview for comparison Operator makes decision which is recorded 10 Goes to Idle About 0 5 second from the car entry a oo setting Next Car triggers the sensor Confidential Power Automation Printed 5 29 2005 J ed Ci a weenie arion auaa GUBE Page 35 of 104 Dawa Sewbalak 13 3 Suggested LPR Procedure ENROLMENT EXISTING EMPLOYEES The existing card owners would have their license plates entered into the existing database and linked to their name facial verification and car details The member s database would be locked allowing only authorised users to add new members When the cardholder vehicle is seen the kcal ink eines ise system will expect the card and facial i verification associated with them also logging them with details of time of entry and name associated the license plate The optional SEE SPEAKER system allows the guards to immediately learn who is who and address each regular parking person by name CASUAL VISITORS When CASUAL VISITORS were KBYAAAIGP E expected their details could be entere
25. operate immediately logging all vehicles and if the number oe me Ali plate is in the database opening the boom E To connect the Facial Verification to the LPR click iy the LPR RE LINK j FANS LPR FRS LPR LINK TO FACIAL PACUME oa ane Facial Front End Power Automation Printed 5 29 2005 Page 48 of 104 Dawa Sewbalak If the software is closed for any reason select the following ICONS to being the LPR and FRS software am m Confidential me e H ARAR PLATE LPR FRONT END FRS FRONT END Adding a user to the LPR database File Edit View Help Select the users ICON by m B gt Hoo o o clicking on the person or by Settings Liwe Users Clear Split To pressing F5 or by selecting EDIT USERS from the ee rep ee x LPR front end screen i Type your password ai Cancel In order to add a user to the LPR system you need to be an administrator with a password Once you have typed in your password you will be allowed access to the MEMBERS DATABSE MANAGER Authorized Members Database Manager View area oa Power Automation Printed 5 29 2005 Se ee se CUBE Page 49 of 104 Dawa Sewbalak A Sequential number within the members list automatic Code The vehicle plate string Used to determine if the recognized vehicle is an authorized member and to optionally open the gate for this member First ee other t
26. owner and employee vehicles would initially need to be registered on the system and an appropriate mechanism for doing this would have to be agreed with the PORT ACCESS CONTROL and the various stakeholders Contractor and supplier vehicles would also need to be registered through a similar process and rules governing their access rights would need to be agreed Processes for visitors and ad hoc entrance would also be agreed and would need to be implemented In Summary The License Plate Recognition and facial capture solution allows Confidential Power Automation Printed 5 29 2005 ee es a aT ee Se HCUBE Page 16 of 104 Dawa Sewbalak Image trigger via a loop on the road Multiple image capture by a high resolution cameras in STEREO front and back License plate finding within the image License plate number reading at an accuracy over 99 95 with multiple cameras used in stereo e Linking of the license plate with a drivers facial template AND OR an IMPRO access card if enhanced security is desired e Linking of the license plate with car colour and shape if required Linking of license plate number with a name and display of this information on ELECTRONIC DISPLAY if present when the vehicle is detected Log of number and image of all vehicles and driver who enter and exit the facility Facial capture and identification on exit if requested TE Ta Image capture of the person from a high resolution colou
27. security guards maintaining accurate databases reviewing identity documents and asking personal questions have proven to be costly stopgap measures that can be defeated by enterprising criminals Compared to other methods of proving identity biometrics are the only tools that can enhance personal privacy and still deliver effective solutions in situations that require confirmation of identity lol rr i Pree Poe Ey E Pe E RE wi Confidential Power Automation Printed 5 29 2005 ee ee ns k a CUBE Page 46 of 104 Dawa Sewbalak 17 LICENSE PLATE RECOGNITION LINKED TO FACIAL VERIFICATION USER MANUAL CONTENTS PAGE Terms and conditions of use Normal Operation Adding a user to the LPR database Restricting access to a user in the LPR database Adding a user to the FRS database Restricting access to a user in the FRS database VIEW HISTORY Access the LPR LOG VIEW HISTORY Access the FRS LOG Terms and conditions of use 1 s provided as a demo UNIT and does not comply with the final product Only users trained by POWER AUTOMATION can use this system Confidential Power Automation Printed 5 29 2005 De Lee es HCUBE Page 47 of 104 Dawa Sewbalak Any DEMO of the equipment can only occur with the permission of POWER AUTOMATION and with POWER AUTOMATION present Confidential NORMAL OPERATION When the PC is started the required software LPR and FRS is automatically started LPR begins to
28. the operator decides on the appropriate action to follow Confidential Power Automation Printed 5 29 2005 TE Oe a cee me LCUBE Page 17 of 104 Dawa Sewbalak ITEM DETAIL No Provided software multiple Databases Pa buamoiene mate arcs ome ene a view closest matches for full details see the user manual below Video capture device eee e Microsoft Windows 2000 Professional Service Pack 3 or If higher e 1000 MHz Pentium 4 Processor e 128 MB RAM e 10GB HDD If required Key amp M Wireless Keyboard amp Mouse If required BNC to RCA cable If required 15 Min standby If required TRAINING a is m 1 Day ON SITE training limited to 6 Perr ay Sy eer ra tS Fars people MAINTENANCE MAINTENANCE ongoing support FREE software updates The I CUBE Facial Identification system is based upon a two tier Client Server architecture The system consists of a single Server application connected to multiple Client applications All data pertaining to images associated text and biometric templates are stored centrally on the I CUBE Facial Identification Server and facial recognition operations may be performed on either Server or Client machines The I CUBE Facial Identification System can store an unlimited number of individuals for use in identification operations The system is set up such that the Server machine contains the Central Server FRS Training Server FRS Database and C
29. to vary significantly over rather short distances It is normal for people new to the system to stare fixedly at the video display during enrollment without exhibiting any movement To achieve optimal performance you should create a situation in which the user is moving their head and walking naturally as well as being exposed to some variance in ambient lighting conditions The system may also be configured to issue verbal commands to the user in order to guide the enroll procedure olan r 3k re Restricting access to a user in the FRS database It is suggested that one does not remove people from the Facial Database but just change the user ID or LICENSE PLATE DETAILS So if the Plate number was NP123 change this to sie eet oe NP123PLUMBER OR NP123LEFT COMPLEX or NPI23GUEST or NP123FIRED etc depending on the reason they are removed from the system PLEASE REMEMBER that this DB is limited to 1000 users the full system is not limited at all Power Automation Printed 5 29 2005 i FEEFEE J EE bet Soa HCUBE M emmi OE a0 53 0 104 VIEW HISTORY Access the FRS LOG In order to view the history file click on Event Viewer E ventviewer The Activity Log Viewer displays the list of operating events that have occurred within the Discovery Client Server installation Provided within log is the time and date in which the event occurred identity of the Discovery Client or Server machine Confidential associate
30. 05 Page 7 of 104 Dawa Sewbalak 4 Abstract An PORT ACCESS CONTROL engages in a constant battle to use technology to enhance their competitive position and detect and prevent crime In order to overcome the problems associated with being at this technological edge Power Automation selects rock solid technology and integrates and supports the technology Implementing the enclosed solutions will allow all role players to gain a competitive advantage in terms of marketing customer relations security and surveillance In order to overcome the problems associated with entry and exit operations where management cannot be present the possession of accurate information regarding what occurs as well as a way to quickly and accurately obtain that information is crucial Power Automation will supply and install the latest digital video technology that will provide a tool that will dramatically change current operations for the better The proposed solution consists of cameras connected to a digital recorder which will allow both real time and recorded view of the entry and exit lanes The entry area in particular will be monitored linking all number plates vehicles and driver seen to the database and video footage The residence plates will trigger the facial verification which if the person matches the facial template in the database will allow the boom to open on entry and departure Visitors and cars pre registered will have the ability t
31. 2 2 1 3 General Panel The facilities provided within the Common panel allow the operator to enable or disable synthesized speech reconnect to a different Discovery Server and provide password protection for program option settings 22 2 1 3 1 Enable Speech By default synthesized speech within the Discovery Client application is not enabled To activate speech check the Enable Speech box Click OK and restart the Discovery application Speech synthesis will now be enabled for all operations such as enroll classification and verification See the Setting Up Speech section later in this chapter for details on customizing speech 22 2 1 3 2 Change Server Name This control is enabled only for Discovery Client machines If this control is checked the Discovery system will request the Server name on the next start up of the application This feature allows the user to connect the Discovery Client to a different Server within the network Confidential Power Automation Printed 5 29 2005 OPP ea aT ee CUBE Page 70 of 104 Dawa Sewbalak 22 2 1 3 3 New Password The operator may enable password protection for the setting of options and pushbutton configurations Password protection is disabled by default Entering characters within this text box and confirming text box enables password protection Password protection may be disabled by clearing the password and confirming text box 22 2 1 4 Restore Defaults lf the system be
32. 2005 Page 74 of 104 Dawa Sewbalak text string i e Could you please register with the receptionist thank you The option for selecting either automated enrollment or speech response is provided within the Enrollment tab 22 2 4 2 General Panel The general panel allows the operator to set the number of closest matches displayed during classification operations and disable or enable speech synthesis for speaking the users name 22 2 4 2 1 Closest Matches This number determines the number of people that will be displayed on the right side of the screen when the Closest Match display is enabled The maximum number of closest matches displayed is five 22 2 4 2 2 Speak Name lf this box is checked the system will speak the individuals name in addition to a preset speech string following a successful classification If unchecked only the speech string without name appended will be generated By default all synthesized speech is deactivated and must be enabled in order for this option to be enabled Refer to the description of the General tab for details on activating speech xi Genma Tracking ImageFiter Eneckmert 22 2 5 Verify Tab vei Owi O Enio Parameter These settings affect the verification one to one a identification operation The option is provided to set Verily Thieshokd EE recognition thresholds confidence lag factor head Faly i orientation boundaries and the distance range applied naet eachd E
33. 4vac Day colour Night B W 5 50mm VF A I Video drive Lense GH 24 Housing IP 68 with heater and wall bracket Loop Controller RELAYS 5VDC 30 MA On site installation and support Remote management software for LPR LPR Software Config Training SUB TOTAL V ideo Qty ha LP A YN O oOo Aa fF OCD DOGO OORA O O 24 24 24 24 24 24 Confidential Power Automation Printed 5 29 2005 Facial Capture camera Lense 3 5mm Push button for facial capture 5 50mm VF A I Video drive Lense UL Mini IR Illuminator 75W 10 Degrees IR illuminator PSU 12Vdc Linear Daynight Switch for IR Illuminator GH 24 Housing IP 68 with heater and wall bracket 24VAC PSU 2A Sundries SUB TOTAL Video Installation Camera installation incuding back box rawbolts focusing and setup RGS9 installed RGS59 Crimp connectors Cable Cabtyre 3 core 1 mm Installed Cable Cabtyre 3 core 2 5mm installed Sundries SUB TOTAL Rack and UPS 2 KVA 2hrs Backup with manual overide 25U 19 rack 800 deep Keyboard tray and keyboard included 3 x modem trays 10 x dedicated power supplies 2 x fans 1 x brush tray Rack preparation Sundries SUB TOTAL Digital Video Recorder Digital recorder Geovision 16 input card IO Card amp Contact Boxes SUB TOTAL Containment and PSU 25mm bosal installed m 204 Box 204 Box mounting kit 20 mm adaptaflex 20 mm adaptaflex connectors Electrical sub db installed Dirty power Electrical sub d
34. After generating or updating a biometric template the Training Server reads the template back into the FRS database via the FRS Central Server There may be multiple FRS Training Servers running simultaneously each connected to the same FRS database through the FRS Central Server PARS Training Ma R Sarem Mare DEH Trang User D When a Training Server is running on a computer the FRS Training Training Type Manager icon will appear in the tray If the Training Server has been loaded but not currently training the icon will be grayed Double click the following icon to display the FRS Training Manager window Tratning Lops The FRS Training Manager window displays the name of the host computer that is running the Training Server the ID of the individual whose template is iio Trairi Type The T mre Ha Cdra pay Cira He r P currently being generated or updated and the type of training that is being performed background vs continuous Background training is performed once immediately following an enrollment or re enrollment operation Continuous training is performed continuously over all users currently enrolled within the system To stop continuous training on the local Training Server click the OFF button In the above example the FRS Training Server is running on computer DEMO and user 81 is currently undergoing continuous training A log displaying the training histo
35. D WARRANTY The Company warrants new goods sold hereunder to be free from defects in materials and workmanship and to be of the kind and quality specified in the proposal for a period of one year from the date of supply The Company makes no warranty whatsoever as to Any goods sold hereunder which have been repaired or altered by anyone other than the Company Goods manufactured by others which may be incorporated with equipment installed or sold hereunder however the manufacturer s warranty for such goods shall be assigned to the purchaser if possible 1 4 1 5 Purchaser and the Company agree that purchaser s sole remedy against the Company and or its suppliers for any defects in the goods sold hereunder whether purchaser s claim arises under the warranty set forth above or otherwise shall be limited to the repair or replacement at the Company s option during normal working hours of any parts FOB the Company s source of the parts The Company shall have no obligation to pay for installation or removal of said parts If goods manufactured or sold by the Company are installed or installation is supervised by the Company or an authorized agent the warranty period shall commence upon completion of installation provided installation is not unreasonably delayed by purchaser in which event the warranty period shall commence when installation could have been completed absent such delays On all other goods the warranty period shal
36. G images from the Discovery System The dialog for configuring the NetAlert Server that provides messages to the Pocket PC is shown below Figure 20 NetAlert Properties dialog The following check boxes are used to enable messages to the Pocket PC 22 2 7 4 1 Enabled Progerties Th eres Saat bee echt ba cere cus at Poke Plc Fiy ce Dekhon Mae fe Enabled Aiai Types Saree Ade mecceges hor the fndiceamg ereentr he Fes imj fe Fae Unbe FAS Dror Log ie Cif ropan fishies Biore FAS Di HTML hiik a CA ropan Fletch ps Bioneer AS Die HTML b r lirip Ci pan Hia a Ea FAS Dae HTML boober janpi C Progean Filer cS yo Eiane FAS Dies CE ces Enables and disables the notification of security alerts The Discovery System will only send messages to the Pocket PC when this box is checked 22 2 7 4 2 Face Classified Sends an alarm to the Pocket PC when a user is successfully classified one to many identification 22 2 7 4 3 Face Unknown Send an alarm to the Pocket PC when a classification operation times out The following filenames must be entered into the text boxes in the lower section of dialog in order for the system to function 22 2 7 4 4 Path to FRS Surveillance Log file Confidential Power Automation Printed 5 29 2005 Page 80 of 104 Dawa Sewbalak The NetAlert Server receives its information from the FRS Discovery log Set the path to the log file here This path must be set before Ne
37. PECOS interface is shown Fe below Figure 19 PECOS Interface Options dialog Piod Sirah ee Merson Line Ween 2 22 2 7 3 1 General Settings 22 2 7 3 1 1 Enabled Checkbox When this checkbox is enabled the extension will send messages to the computer PECOS server specified by Server Address and Port using the version protocol specified in the Protocol Settings 22 2 7 3 2 Protocol Settings 22 2 7 3 2 1 Use Version 1 When this option is selected both READY and MATCH messages will be sent to the machine specified by Server Address and Port PDA Application Settings are only available with Version 1 22 2 7 3 2 2 Use Version 2 When this option is selected pseudo XML messages will be sent to the machine specified by Server Address and Port These messages will contain detailed information regarding User Location Time and alarm type A snapshot is also embedded in the Version 2 message PDA Application Settings are not available with Version 2 22 2 7 3 2 3 Server Address IP Address or UPC name of the machine to which all messages will be sent This should be the machine on which the PECOS server resides If a UPC name is used it must be registered with your network s DNS IP Addresses are generally more reliable 22 2 7 3 2 4 Port Port on the machine that will receive messages The Port number must match between PECOS system server and this extension All external systems connect to PECOS on the same P
38. Power Automation Printed 5 29 2005 WTF et Le ns oe CUBE Page 15 of 104 Dawa Sewbalak When visitors are scheduled to arrive security can be contacted to enter the registration of the vehicle into the system and when the car arrives access is allowed and a welcome message displayed on the visual display if supplied Access for visitors can be defined for the period that they will be there and while the record will still in the DB it will be in a status that would recognize but not allow access Facial recognition could be added to this to further enhance the security access control system to log all drivers with the vehicles Facial verification would ensure that there is a match between vehicle and designated driver It will be possible to link multiple drivers to multiple vehicles so the system would not be restrictive from that perspective The idea is to remove much of the responsibility of assessing vehicles entering to dealing with the anomalies and allow normal authorized vehicles to proceed without delay COMPONENTS Typical SeeLane Configuration a Remote Database The main system would be installed on a server controlling the process This would house the database as well as the application that handles the analysis It would be linked to a trigger mechanism either a loop at the entrance or the presentation of the proximity tag the boom and the various cameras OPERATIONAL ISSUES All PORT ACCESS CONTROL personnel
39. TE RMR ATP IHAN mea Daw CUBE Page 1 of 104 Dawa Sewbalak Proposal for PORT for ADVANCED VEHICLE ACCESS CONTROL SOLUTIONS Incorporating LICENSE PLATE RECOGNITION DIGITAL RECORDING and Biometric Facial Identification VERIFICATION SOLUTIONS Compiled Supplied by Dawa Sewbalak m Sunday 29 May 2005 In conjunction with l Cube Confidential Power Automation Printed 5 29 2005 Page 2 of 104 Dawa Sewbalak TABLE OF CONTENTS 1 Confidentiality Clause cccecccceecccceeeeeceececaeeeeeseeeeseeesseeeeesneees 4 2 SGI AU OU ata cutser ews cuten E EE dbo ssucusvossuan aera E EEEE 5 3 ACKNOWIEACGEMENMS ccccccceececeeecsececeeeceeeceueeseeeseeeesueeseeeeeesaes 6 4 PS AU csc cess cee E E E 7 5 Description Of REQUIFEMENTL ccccccecceeeeeeceeeeeeseeeeeeceeeneeaess 12 6 Issues for Consideration ceccceececeeccceeceteeceseeeeseeeeeeetaneeeaes 13 7 ONION Propos CG sagecatecece A sad cacewe se E E 14 7 1 OOM DESIO issii see secs secede acest anaana EA ea EE oR ER 14 7 1 1 Vehicular MAMI G von cnceectecetcsesciecaccsciiase isona NNT 14 7 2 Face Recognition System DeSCTIPtiON cccceeceeceeeceeseeeeeceeeeeeseeeeeeseeeeesaaeeeesees 16 8 FACIAL VERIFICATION Equipment Requirement 18 8 1 HARDWARE estan pce de vescanieet pneteeneeecncaceann secede seccenesetaisecadeeceaeatecet sue sodecensenaceansetaeees 18 8 2 Additional REMS petisds cance cincs at
40. Windows 98 NT 2000 e See Car DLL which is used to analyse the images and extract license plate string e Camera Illumination unit to capture the images See Car Head LPR camera and illumination unit e a Frame Grabber which captures the images from the camera units handles 1 6 lanes e 1 O card input output board with multiple I O discrete lines This board supports the sensors illumination control and optional gate open signal It is connected via a cable to a terminal interface board with easy connections and indicator lights e Sensors to indicate the presence of the car a sensor for each lane e See Lane The See Lane Windows application interfaces the hardware elements frame grabber camera illumination unit s IO card and sensor It controls the illumination reads the video inputs and passes the images to the DLL in order to obtain the recognition results The application displays the image and recognition results It then exports the results using serial communication messages or disk files Its man machine interface supports on line setting control which can easily adapt the application to various types of configurations x ee The camera is triggered by the loop in the road SeeParkClient is an application that is used to monitor a parking lot secured area with a specified number of entrances and exits This application records the entrances and exits in a simple flat database and uses the information to match
41. ablished as a fraction of the height of the video frame 22 2 3 2 Mask Thresholds Panel The panel entitled Mask Thresholds panel pertains to a series of masks that are applied over the video frame sectioned into 16 x 16 regions Each region of the video frame is evaluated separately for variance frame to frame movement and color content Thresholds are applied to these measurements in order to mask out regions of the video frame from the tracking subsystem Proper use of masks can significantly improve tracking speed and performance It should also be noted that the effect of the masks may be viewed by clicking on the View Show Masks All menu items Descriptions of the parameters contained within the Mask Thresholds panel are provided below 22 2 3 2 1 Variance The variance measurement within each region is normalized by variance across the entire video frame This set point is used when the variation over time in overall lighting intensity is large If relative pixel intensity variance is less than this threshold the region will be masked out 22 2 3 2 2 Abs Variance Similar to the Variance threshold however the threshold for masking out regions of the video frame based upon an absolute measure of pixel intensity variance If intensity variance is less than this threshold the region will be masked out 22 2 3 2 3 Movement Sets a difference threshold for frame to frame pixel intensity This value is used to detect motion If th
42. accept the enrolled images The Accept Images window provides the operator the option to remove facial images of poor quality or images of people who may have been incorrectly tracked during the enrollment operation ocepi images Total 100 Enroll Compleied Ed Figure 6 Accept Images window is shown at the end of enroll The operator may scroll through the collected images using the and gt buttons Single images may be removed by selecting the image and clicking the Remove button If the operator is satisfied with the enrolled images click the OK button to accept images and allow the system to begin generating the biometric template for that user The system will collect images faster if the user being enrolled provides different expressions at different head angles lighting conditions and at varying distances from the camera It is normal for users new to the system to stare fixedly at the video display To achieve optimal performance you must create a situation in which the user is moving their head and walking to create varying lighting conditions The operator should instruct the user to move naturally and or walk towards the camera The system may also be set to issue verbal commands to the user in order to Guide the enroll procedure see the Setting Up Speech section at the end of this chapter 22 1 2 6 Show Options Click the Options button or select Options from the Control menu to display the A Opti
43. ader via DDE TCP IP or R8232 The biometric system would link the VISITORS info to the license plate and if required a parking ticket and allow the boom to open EXIT USE EMPLOYEES The existing long term employees would drive up to the exit gate The LPR system would capture the plate and pass this information to the facial verification and DVR system If the facial verification matches the enrolled template and the parking ticket is correct the boom will open The DVR system would then link the vehicle license plate to the driver and car seen on the cameras VISITORS All VISITORS exiting would be identified by LPR which would transmit the details to facial verification reader via DDE TCP IP or RS232 If the license plate the VEHICLE ticket info and the biometric ticket issued match the boom would open If not the guard would investigate All video images will be digitally recorded Video images can also be transmitted to the main office if required Transmission of the camera images to a remote monitoring control room will be possible 13 4 Schedule Of Equipment Specifications See attached EXCEL file Confidential Power Automation Printed 5 29 2005 FEEFEE J EE Z HCUBE 14 LPR Equipment Quote TO PROVIDE FIXED BLACK amp WHITE CAMERAS WITH 16 MM FIXED IRIS LENSES AND BRACKETS SeeCarHead s Integrated Camera lllumination unit s housed in a weatherproof enclosure Power supply for SeeCarHead units
44. ancel specific record in the view area Clicking _ cce on the Edit button will open a separate Last name Dudley Help window same window as Add new member with the record details It is possible to edit this record Another way to edit a record is double clicking on the record in the view area Authorized gates Gated Gate2 Delete Enable access to all gates Delete a record This button is disabled It IS suggested you do not delete plates but disable access to gates hence you will always keep a record of all visitors and cars allowed access a a Lie i E Lines Efika laeg ives Two lane front end Restricting a Ca access to a user in the LPR database xj It is suggested you do not delete plates but disable access to gates hence you Vehicle code I Vehicle code npr 2 First Mame Cancel First Marne Barry Cancel Last name Help Last name Dudley Help Authorized gates Authorized gates M Gate 1 Gate 2 Gate 4 Gate 2 J Enable access to all gates Enable access to all gates Confidential Power Automation Printed 5 29 2005 MTE PFE Pe Tp cae ee CUBE Confidential Page 51 of 104 will always keep a record of all visitors and cars allowed access Adding a user to the FRS database Enroll When adding a person the FRS system you HAVE to keep the SAME name and LICENSE PLATE number To enrol a person click the enro
45. awa Sewbalak remains activated is adjusted by entering an integer value into the textbox located to the right of the relay contact selection checkbox 22 2 7 2 Cardac Wiegand Serial The Cardac Wiegand Serial extension provides an interface for any Wiegand input device typically a magnetic ID card or keypad Lardac Wiegand Extension On swipe of an ID card or input from any compatible Wiegand device the system will initiate a verify operation If the user has been registered but not enrolled the system will automatically initiate the enrollment procedure See the Enrollment Options for Pasho details D dhay Parhough i Fani on ep onky Fania all Ragiem carck C Have Penh e Enhed Diii To set the options for the Wiegand Extension select Cardac Wiegand Serial Extension from the Extensions list and click Properties The Options dialog will appear C Figure 18 Cardac Wiegand Extension dialog 22 2 7 2 1 Enabled Enables the Extension if checked Disables the Extension if not checked 22 2 7 2 2 Port Specifies the COM port connected to Wiegand Serial Interface device 22 2 7 2 3 Passthrough In all cases when a Wiegand device is activated through card swipe or keypad entry and the user associated with the ID number is both registered and enrolled within the system a verification operation is initiated 22 2 7 2 3 1 Always Passthough The ID number is passed through the system to be received by an external Wiega
46. b UPS with 16way box and CB s Page 23 of 104 Dawa Sewbalak OV OO NN O amp O 24 24 24 24 24 24 24 24 24 24 24 24 Confidential Power Automation Printed 5 29 2005 TL rE ETF AFETE E 1p j dP a Se HCUBE Page 24 of 104 Sundries Electronic Display Display Board software Display Board exluding installation On site installation and support WIRELESS COMMUNICATIONS Base Station Connector to base station PC Board Install Facial Identification Software XP amp 2000 to compare Facial images resulting in 12 closest matchs List Price for FRS Discovery Server software List price for FRS Client Application DB of less than 10 000 Contingency and Project Management MAINTENANCE Based on a 24 hrs turnaround per month Totals Confidential Power Automation Dawa Sewbalak 1 SUB TOTAL 3 3 l SUB TOTAL 1 6 6 SUB TOTAL l l 6 l SUB TOTAL 12 12 Printed 5 29 2005 KTF Pe a j H ee Se CUBE Page 25 of 104 Dawa Sewbalak Conditions e All prices are quoted in US DOLLARS and are subject to change e Proposal is valid for 90 days after which the pricing and product availability may well have changed e Neither installation nor commissioning has been quoted due to the unfamiliarity with the environment e Maintenance will be quoted separately 9 1 Acceptance Customer Acceptance CUSTOMER VENDOR POWER AUTOMATION Signature Signature Name Name Title Titl
47. biin T ppa SOL Serve e Enter the name of the computer hostin the database in the Staines Server box cane e Enter the name of the database itself in the Database Name box e Select the Database Type from the drop down list e Click OK Databeres Meme FASDE The Database window will close and the database name will appear in the Database box in the FRS Server Manager window Checking the Authentication box will force the user to enter a username and password each time that a connection to the database is attempted i e each time FRS Central Server or FRS Training Server is started Be aware that the username and password can be different for each database The default username is frsuser and the default password is password Right clicking on the FRS Server Manager tray icon will display a popup menu that allows you to show and hide the FRS Server Manager window and shutdown and restart the Server Selecting exit will not only hide the FRS Server Manager window but stop the FRS Central Server itself 22 10FRS Training Server Manager The FRS Training Server continuously updates biometric templates using the images that as are enrolled for each user Confidential Power Automation Printed 5 29 2005 MTF RFETE D A1 es LCUBE Page 92 of 104 Dawa Sewbalak The Training Server s receive facial images from the FRS database this transport layer is controlled through the FRS Central Server
48. d prior to arrival so that when they arrived they can be linked to an access card If they were not on the system the driver would drive up to the entrance a picture of the plate would be captured OCR would occur after which the driver would do facial verification linking the printed ticket to the license plate The natural process of e mailing calling or filling a simple completed VISITOR application document directly to the system means that when a casual VISITOR arrived at the counter their license number is automatically entered into the custom VISITOR system and their authorised record would appear Access to the correct areas with the correct expiry is C automatically issued E D ecn j ENTRY USE Te ene E oh better ff EMPLOYEES The existing long term employees would drive up to the gate any of X The LPR system would capture the plate and if the plate were in the database the LPR system would link the plate to a card if this matched this would then open the gate The DVR system would then link the vehicle license plate to the drivers face and the vehicle If there was any concern the guard would investigate past events to see the driver and car matched the plates Confidential Power Automation Printed 5 29 2005 Ue Ud HCUBE Page 36 of 104 Dawa Sewbalak VISITORS Pre registered VISITORS would be identified by LPR which would transmit the details to the facial verification re
49. d user name if applicable and the type of event The Activity Log also provides an image captured ae from the Discovery application at the time that the event occurred aelbdi The events that are recorded and displayed within this log are as follows Enroll Failure Enroll Success f l BTT FESJITEST I TETI TTI Erara DEG mE g THT roan EPITETI a EJI E eee HR F Reenroll Verify Failure Verify Success Classify Failure Classify Success i TTT T Pees pene See eee baa Multiple activity logs may be generated and viewed using the multiple document interface provided within the Activity Log Viewer Each view may be filtered to list events by event time and date computer name user name and event type Any report contained within the Activity Log Viewer may be saved and restored from a binary file The report may be also saved to a text file for import and processing by third party software The list of events provided within a window of the Activity Log Viewer can be saved to a text file for later processing Note that the associated JPEG images are not exported when saving the selected activity log to a text file To save an activity log to a text file perform the following operations Select the window containing the activity log that is to be saved Select the Save Text File menu item from the File main menu A standard File Save dialog will appear
50. during Verification i Oo Track Taen jrd i Figure 14 Options window Verify tab 22 2 5 1 Parameters Panel A descriptions of the various parameters that may be adjusted for the verify operation are as follows 22 2 5 1 1 Threshold The output value recognition confidence generated by biometric templates range from E ee 1 to 1 with 1 indicating high confidence in recognition This threshold establishes the set point above which an individual is positively identified 22 2 5 1 2 Frame Lag The confidence values generated across all biometric templates are lagged by the video capture frame count This value establishes the degree of digital lag applied to the recognition confidence values when the system is operating in verify mode Confidential Power Automation Printed 5 29 2005 Te ee Le CUBE Page 75 of 104 Dawa Sewbalak 22 2 5 1 3 Timeout seconds Sets the maximum number of seconds provided for the verify operation The corresponding timer is incremented only while tracking is active In the event that the individual is not verified within this cumulative time interval the system registers a verify failure and generates the pre set response i e synthesized speech 22 2 5 1 4 Track Timeout seconds Sets an absolute timeout period for the verify operation regardless of whether tracking is active i e timeout condition in the event that the user walks away from the camera prior to being verified I
51. e Date Date Specific considerations Please sign and fax this page to POWER AUTOMATION Dawa Sewbalak Power Automation E Mail power auto intnet mu 9 2 Standard Terms and Conditions a 7 ii i Available on Request LLETETE L et Ba MERCEKRIWP ist EO PETET 2 a E EEI DE Confidential Power Automation Printed 5 29 2005 i ip EE a Ba isa m m TH dee m E im Ta T i ima i z wa B iE m _ a i il a THe a al PAE TE eo Lo TELIOT el hee ee ee ns k i Page 26 of 104 Dawa Sewbalak 10 Rental Option 10 1 BENEFITS OF POWER AUTOMATION RENTAL Many of our customers ask whether it is better to buy Cash or Rental However when buying high tech electronic equipment one has to remember that with today s pace of development that obsolescence needs to be taken into account Because the equipment is computerized and networked we are totally dependent on the likes of Microsoft This means that the average lifespan of computerized equipment is 3 Years The receiver of Revenue has accepted this hence all computer equipment is depreciated over a 3 year period The present trend for the increasing cost of importing such technology is averaging between 20 30 per annum and does not make it financially viable to purchase a new system every three to four years This is a conservative estimate and is due to a number of factors
52. e captured allowing for immediate recall and review The LPR system allows the driver to be identified immediately if the guard has any cause for concern FIGURE Currently existing parking LPR solution Sample of system accuracy where read seq is set to 100 CHEP LPR RECOGNITION RESULTS Confirmed Recognition 99 90 Recognition per event Vehicle driven out while 2nd vehicle on loop 0 2 TOTAL 100 0 13 1 Full description of system topology offered E Number 18136 za Field LPR engines connected to the LPR cameras by frame grabbers 6 inputs capture the images and process the information from the Parking Garage the result of license plate recognition and facial and vehicle colour capture This information and the images are transmitted to the central server by the network immediately being available for review and reaction by any of operators at the operator stations Confidential Power Automation Printed 5 29 2005 meme e ETF AFETE Ap TPA k Se CUBE Page 32 of 104 Each field system can operate totally independently and will update the central server as soon as this becomes available As such the HOT list is downloaded to each field PC so no illegal unwanted vehicles might use the weighbridge os System Architecture The License Plate recognition product is a turn key system comprises P Mems et eum of the following elements e aPC Pentium running
53. e attached solution This solution allows management to BE EVERY WHERE AT ONCE and focus on Productivity improvement Loss control Vehicle monitoring Safety management Ensures that staff and managers spend time where they are needed most SSSss This Digital Management Solution will be implemented into your business to help you better manage your business and in the long run improve your profits through increased productivity levels and better work practices With POWER AUTOMATION Digital Management Solutions you will have better control over access control visitors deliveries contractors and staff ever cameras ave your record JANI MEYER succensfid thatthe coptial city s cloaedtircsit tnlevision cam fees bs in the ersten the specdistors are nothing comers euirrotes siba ii o Hs memory banik Rip ut Eee inara com Inartit remembered aml recog needs Hf Whal mete the syaani npari ba that anci implemented tH te gonstanth Warning Oiice a Confidential Power Automation Printed 5 29 2005 z MERE PER iT p eT HCUBE Page 10 of 104 Dawa Sewbalak Implementing our digital system throughout your premises will give you a visual audit trail of cars giving you much more material to work with when making managerial decisions The system doubles as a security and Visual Management System and since it s an open architecture system other compatible systems can be
54. e frame to frame difference in pixel intensity is less than this threshold the region will be masked out 22 2 3 2 4 Face Color Regions of the image may be masked out based upon color content This value sets the mask threshold for skin tone The color scoring method has been tuned to recognize skin tone among all ethnic types The valid range for this parameter is 1 to 1 Lower values will Cause a greater region of the video frame to be masked out Confidential Power Automation Printed 5 29 2005 ETF ee es a aT ee Se CUBE Page 73 of 104 Dawa Sewbalak 22 2 3 2 5 Advanced The above masks may be applied in various combinations GEEEETES 55 using an 8 frame cyclical schedule This schedule may be Fem 12345 6 modified by clicking on the Advanced pushbutton which assnes i lv fe It fe EY enables the following screen Mower CRCOPECeECe Figure 12 Tracking Masks dialog Feats CIRRCCERE The masking conditions that are checked along each column are combined using a logical OR relationship 22 2 4 Classify Tab These settings affect the classification one to many identification operation The option is provided to set recognition thresholds confidence lag factor and the classification timeout value Figure 13 Options window Classify tab Parameters Panel OOO hhh Descriptions of the various parameters that may ee O a ee a a be adjusted for the classify operation are as Veri Ces Extensions e f
55. e goods the repair and maintenance of any faulty goods to the original operational condition and the recalibration and re commissioning of the affected goods promptly in order to ensure the downtime is kept to an absolute minimum d The maintenance services include the cost of all parts and consumables unless such parts or consumables are specifically excluded Should you wish to consider our Silver or Platinum Maintenance plans please feel free to contact me Confidential Power Automation Printed 5 29 2005 Ue Ud CUBE Page 57 of 104 Dawa Sewbalak 19 Equipment Requirement The performance of the PC and camera equipment is subject to a 220 volt power source at each camera position is to be provided If required we can arrange an electrician at an additional cost The CCTV signal from the cameras requires fibre to get back to the control room if more than 75M from the PC ui tet Tia Petr EEE i H MIO GATE ESTATE HOME OWNERS ASSOCIATION Confidential Power Automation Printed 5 29 2005 atone tia ce GUBE Page 58 of 104 Dawa Sewbalak 20 Digital Recorder Integration Both the LPR and facial identification incorporate into existing DVR systems allowing the use of existing infrastructure turning data into information tim L km Fo OL Some tes la fe Wi i kau Pom i i ea Pa iia a fen Fam i f bimi a E fom Fam 5 i fo So PH LI rd fom e LY Cem l
56. e that people were who they claimed to be Modern computer based biometric systems are employed for identification who is this person in which a subject s identity is determined by comparing a measured biometric against a database of stored records a one to many comparison Chip or reader embedded in turnstile Voice recognition Microphone S Facial recognition Video camera surveillance camera single mage camera Acquisition devices associated with biometric technology Despite vendor claims there is no ideal biometric technology although examples of successful uses opisy Analyus exist Facial recognition a technology that has lt gained ground in recent years thanks to the falling aa price of computer power It works by analysing a video image or photograph and identifying the positions of several dozen fixed nodal points on a ih AARAA person s face These nodal points mostly between et E a the forehead and the upper lip are only slightly en te affected by expression or the presence of facial F ee iiaa diate Artinsacan i i Fingin hair Facial recognition is becoming more i widespread because it can exploit existing cameras and existing databases of facial images from driving licences and passports Exclusion techniques based on BIOMETRICS have some serious technological advantages If a single positive identification can prevent a theft then the sooner one begins to use the technolo
57. e to compute whether a biometric sample and template are a match See also Artificial Neural Network Attempt The submission of a biometric sample to a biometric system for identification or verification A biometric system may allow more than one attempt to identify or verify Authentication Alternative term for Verification Automatic ID Auto ID An umbrella term for any biometric system or other security technology that uses automatic means to check identity This applies to both one to one verification and one to many identification Behavioural Biometric A biometric which is characterised by a behavioural trait that is learnt and acquired over time rather than a physiological characteristic However physiological elements may influence the monitored behaviour Biometric A measurable physical characteristic or personal behavioural trait used to recognise the identity or verify the claimed identity of an enrolee Biometric Application The use to which a biometric system is put Biometric Data The information extracted from the biometric sample and used either to build a reference template template data or to compare against a previously created reference template comparison data Biometric Engine The software element of the biometric system which processes biometric data during the stages of enrolment capture extraction and comparison Biometric Device The part of a biometric system containing the sensor t
58. ect Save from the Fi e menu A standard File dialog will appear e Enter the name of the file used to save the activity log If you choose an existing file it will be overwritten e Click OK The activity log will be saved to the specified file Activity log files have the log extension 22 5 5 Loading the Activity Log The data from an activity log file can be loaded into an existing view To load an activity log perform the following operations e Select the view that will receive the activity log data Note that the loaded data will overwrite the current view s data e Select Load from the Fi e menu A standard File dialog will appear e Locate the activity log file to view Activity log files have the log extension Button Configuration x e Click OK The activity log file will be loaded into a new Buhan T pres view FF Track buan 22 6 Modifying Control Button Configuration E The buttons available on the control bar Track Classify Verify Add 5 Pe Weely bukon Closest and Options may be hidden to limit the functionality of the Cass marha bim Discovery application he did buton To hide selected buttons on the control bar F Oni Hira Bullen Click the View menu W Ful graan baion Select Button Configuration from the View menu The Button Configuration dialog will appear Figure 30 Button Configuration dialog Select the buttons that will appear on the control bar Click OK The Button Configuration dia
59. ectives Merchandise Alarms Fitling Room Attendants Wiorcae Electronio Security Tags POS Esorption Gased OOTY interface Timed Entry Sates VendorSounce Acouto Magnedio Tagging Cbeenation Booths Wendor Source RF Tagging RFC Merchandse Sariy Tags gW 10 20 3 40 50 GOA 70W 80 90 100 Overall Usage 3 Figure 1 Use of CCTV systems is on the increase of this information is the key to effective decision making POWER AUTOMATION provides the artificial intelligence on a digital platform to turn Images to Intelligence Customised application specific algorithms provide real time processing detection storage and rules based decision support systems of pre configured events with the added advantage of being able to artificially interrogate the recorded database In Confidential Power Automation Printed 5 29 2005 i Ta ra vmnm aac CUBE Page 11 of 104 Dawa Sewbalak essence these systems automate security process and inspections within an operation reducing manpower and the inherent human failure A 12 month guarantee is applicable in respect of all equipment installed and workmanship A separate maintenance contract can be negotiated for the balance of the rental period after the expiration of the guarantee period The system requires either 12V or 220 volt plug points and sufficient light for the cameras although the cameras quoted on have a low light specification All training to the relative users
60. ed 5 29 2005 6 Tn ee 1p eT ee Se HCUBE Page 13 of 104 Dawa Sewbalak Issues for Consideration There are a few points that should be borne in mind during the compilation review and assessment of this proposal and the subsequent manner in which this will progress should your decision be favourable to move forward This proposal outlines what can be done in order to address the stated requirements and does not delve into too much detail Although much of the detail is supplied in the Appendix The detail regarding the actual implementation process is not addressed but is estimated in the Financial Considerations section Our philosophy for solution crafting highlights the fact that all solutions consist of 4 primary elements people process technology and information The technology aspect is addressed in this proposal The other aspects are normally addressed in the project scoping workshop which will typically commence once there is agreement to proceed The success of implementations of this nature is dependent on a formalised project approach where expectations and deliverables are clearly expressed recorded and agreed Ensuring open dialogue prior to implementation allows for all parties to ensure that all parties agree Scope management is also crucial to ensure proper implementation We all know what assumptions do and are A medium level project schedule with project report will be supplied after the project kick off ses
61. el BE Page 5 of 104 2 Declaration This research is done specifically for PORT ACCESS CONTROL CLIENT Date Sunday 29 May 2005 The opinions expressed in this document are the views of the authors alone and do not necessarily reflect those of the views of DECALINK LTDA management employees or any other party Numerous assumptions have been made and many of these would have to be validated at the commencement of the project Dawa Sewbalak Dawa Sewbalak Power Automation E Mail power auto intnet mu BUEN DETEN es em m a Poa 2 PPIPIVRTRTETT leg Perel li ai me BELET i THTRTLITATRA TUTTI EGET LATENT ET aint rti i T EREHE Fidel eee TETETIS ee MERCH RWI fel oer Liu a le Confidential Power Automation Printed 5 29 2005 Ue Ud HCUBE Page 6 of 104 Dawa Sewbalak 3 Acknowledgements We would like to thank the following for assistance Barry T DUDLEY I CUBE I3 Integrated Intelligent Imaging MBA IT MSc Image Analysis BSc Brewing BSc Hons Waste Technology LICENSE PLATE RECOGNITION AND FACIAL IDENTIFICATION SOLUTIONS http www i cube co za Cell 27 0 82 562 8225 MADADENI PH 27 0 31 764 3077 82 Kloof Falls Rd Fax 031 7643077 Kloof Durban Kwa Zulu Natal 3610 South Africa E mail info Cube co za INTEGRATED INTELLIGENT IMAGING CUBE Enrali F A ir Ww decalink cl MKF359GP Confidential Power Automation Printed 5 29 20
62. ental which reflects as a liability on the customers balance sheet the POWER AUTOMATION Rental is a pure off balance sheet operating expense which is 100 tax deductible This alleviates the admin burden of financial asset management and depreciation and furthermore frees up much needed capital and existing PORT ACCESS CONTROL facilities for better use within your business Quote from the September 2002 issue of UPFRONT lt Computer equipment in particular is the most rapid depreciating asset in South Africa Market used equipment and it becomes understandable why most financial institutions attach a zero value to all used digital equipment Confidential Power Automation Printed 5 29 2005 ETF ee es Ap TPHA ed Se HCUBE Page 27 of 104 Dawa Sewbalak 11 Conclusion POWER AUTOMATION will supply and install the latest digital video technology including advanced license plate recognition linked to biometric facial verification identification and recognition The latest LPR technology will be provided as a tool to verify entry and possibly identify an unknown person if they appear in the database We believe that the POWER AUTOMATION solutions will grab the attention of all role players leading to the focus ON Ems Hisra productivity damage reduction and increased information flow allowing a safer environment We deliver ROIl driven solutions that leverage video and control technologies to increase profit by
63. epository of information typically a database This process requires a search through a database and the search can be sequential or can be built on certain criteria to ensure more effective lookup It is a one to many review Verification is the process whereby an item for identification is presented either a license plate PIN proximity card or ID card and an additional criteria in this case facial is used to validate that the car presenting the item is actually who they are claiming to be This is a one to one review Each of these options are possible with the technology proposed but each requires a different approach to the actual implementation and the associated processes 7 1 1 Vehicular Traffic This specific system will allow for the capturing and analysis of images of licence plates front and rear which would also cater for trailers to ensure an accurate record of vehicles entering and exiting the EDUCATIONAL FACILITY Each vehicle could be classified as employee owner visitor contractor Supplier etc Rules for access could be defined for each vehicle and access could be granted accordingly The system proposed would analyse the licence plate and the colour and shape of the vehicle for the match and would then operate the DOOM Thal eog ganm i reii ie n e mate ie oe lf there is a discrepancy then _ the security personnel will be alerted and appropriate intervention will then take place Confidential
64. er Manager window will appear The FRS Server Manager displays the name of the computer on which the FRS Central Server is running The figure above shows the FRS Central Server is running on DEMO The status of the FRS Central Server is also shown While the Server is running the status will be Started The FRS Server Manager also displays the number of Client Computers currently connected to the Server the number of Training Servers used to generate the biometric templates 1 machine by default in the case of the Discovery System and the FRS database currently in use Confidential Power Automation Printed 5 29 2005 Se es ee Page 90 of 104 Dawa Sewbalak From the FRS Server Manager window you may shutdown and restart the FRS Central Server view currently connected Client computers add FRS Training Servers or select the FRS database to which you wish to establish a connection Clicking on the X button in the top right of the FRS Server Manager window will hide the window but does not stop the FRS Central Server 22 9 1 Shutting Down the FRS Central Server To Shutdown the FRS Central Server click on the Shutdown button This will stop the Server s operations but not close the FRS Server Manager window You can restart the FRS Central Server later The status will read Server Down 22 9 2 Restarting the FRS Central Server To restart the FRS Central Server after you have shut it down click on the
65. erform verification or classification operations reliably Enrollment of static images may be performed in conjunction with the standard video enrollment procedure To perform an enrollment from r j pon mam static images select Enroll Static i Images from the Database menu Figure 26 Enroll Static Images window 22 4 1 1 Enrolling a New User from a Static Image A new user may be enrolled from LN ors a static image The system will epee ch Tet automatically assign the new users name and card number Ce _ a j based on the filename of the JPEG The JPEG file name must have the following format LASTNAME _ FIRSTNAME _ CARDID Jjpg Confidential Power Automation Printed 5 29 2005 PL bo a ee ee ns k a HCUBE Page 85 of 104 Dawa Sewbalak For instance a JPEG file with the name Doe_John _25463 will create a user named John Doe with ID number 25463 To enroll a new user from a static image Select From File Selection Select the JPEG file from the file list located in the upper left corner of the dialog Select New User within the user list The JPEG image will appear in the image box The person s face will be tracked automatically If no tracking box appears around the face click the Retrack Face button e When the face is located in the image click the Start Enroll button 22 4 1 2 Adding a Static Image to an Existing User You may add static images to the training set
66. ess since they change Micmkmi Crass Meat over time and differs across cultures By adhering to applicable best practices even those technologies more capable of being misused primarily facial recognition and gt fingerprint can be deployed in ber ta el sm tY a privacy sympathetic fashion m Nitpg eile 0D ipg BioPrivacy Best Practices 2003 Available online at http www ibqgweb com reports public reports privacy best practices html The use of the information gathered for exclusion purposes needs to be weighed against the possible use of the information Fingerprint face and iris have the highest privacy risk It is essential that lS A a E Confidential Power Automation Printed 5 29 2005 MEREN PED aT aT ee CUBE Page 45 of 104 Dawa Sewbalak appropriate protection should be in place to ensure the technology is not misused Mc Cullagh D 2003 Self reporting data would be wrapped in software or digital watermarks that guard against misuse of private information by tracking who has used the data and where they have been moved Roush 2003 The manner is which proper protection occurs is beyond the scope of this study Identity theft using stolen credit cards phoney cheques and other impostor scams to steal is on the increase Vijayan 2003 Until recently the only way to way to attack the problem has been to add expensive screening and administration procedures However steps such as hiring
67. events using the POF string To share the results pass the remote data to the centers an additional utility SeeData is used to pass the recognition results and copy the image files to the Center The product runs on basis of TCP IP and sends the data across the network It is consisted of two parts one running on each remote unit as a background application while the other part runs on the Center Server also as background application The product can also send back commands from the Center to any remote lane in order to activate a sensor or command to open a gate Confidential Power Automation Printed 5 29 2005 OE TT ee CUBE Page 33 of 104 Dawa Sewbalak To simplify the maintenance of our products our SeeMonitor tool monitors the operation of cluster of SeeLane and provides a quick view of the operation in green yellow red status The status is determined by automatically checking the Windows application event logs where our products log information warnings and errors These events include possible failure of hardware units application problems and decrease of recognition results under acceptable levels SeeMonitor also provides various performance graphs for each remote unit and a multiple view of the units These graphs assist in the fine tuning of the system In the following example the graph of the horizontal center of the ID marking is shown as a histogram which can be used to verify the correct settings The gra
68. ewing an event Double click on a event number It will show the image of that event You can also browse with Previous or Next buttons SeeLaneClient The main view displays a list to which items are added for each new data item received from the SeeLane server all the data will be written to file Each row shows another recognition result as reported by the DDE message coming from SeeLane The SeeLaneClient can also send 2 kind of triggers to the Seelane by the DDE server message Open Gate trigger Incoming vehicles Trigger HELP AND ASSISTANCE PLEASE CONTACT Barry T Dudley on 082 562 8225 or LPR I Cube co za Power Automation Printed 5 29 2005 Ue Ul a HCUBE Page 56 of 104 Dawa Sewbalak 18 Installation Quote e Bosal conduit bnc connectors cabling labour excluding any trenching electrical work and poles e Commissioning and training THE MAINTENANCE SERVICE GOLD a The provision of corrective maintenance which means the rendering of services for diagnosing the breakdown of the goods and the subsequent actions necessary to restore the goods to their corrective function b The provision of preventative maintenance which means the observation of the goods with the intention of identifying minor breakdown or deterioration of the goods and the subsequent actions to restore them to their correct functional and operational state c Ensuring the continuous and uninterrupted operation of th
69. ext of 32 characters alent maximum Vehicle code Used to identify the authorized member First Mame Last Last name OK Family name or any other text of 32 characters maximum lv Gate 1 M Gate 2 Used to identify the authorized member Authorized gates J Enable access to all gates Gates Specifies the gate number that the member can pass through when recognized Used to limit the rights of access to some vehicles Range 1 to Maximum licensed lanes Default All gates which grants access for authorized members to all gates One click on the top of each column will sort the list according to the subject for example click on sequential number will change the sort order from increasing to decreasing and vice versa Command buttons area Add Add a record to the members list A separate window will be open with the following fields Vehicle Code The license plate as seen on the vehicle without spaces The First name of the driver The last name of the driver Gate one 1 is the entry lane and Gate 2 two is the EXIT lane Find Find a record according to one or more of the following Vehicle Code First name or Last name A separate window will be open Confidential Power Automation Printed 5 29 2005 OE aT GT ee CUBE Page 50 of 104 Dawa Sewbalak Edit Edit a record This button is disabled It 500S coas Gi 7 can be enabled only by clicking on a Sheet Bar va l l Soak y C
70. f the system is unable to verify the face in less than the specified interval default 20 seconds a failed verify attempt is SS recorded Yei Oei Extensions ene i Fl 22 2 6 Image Filter Tab Genel Tracking Image Fier Ervolnent Facial images captured during enrollment verification Parameters or classification operations may be filtered in order to Hinman ice improve the quality of training data and improve Heed Fotion degec 2 recognition accuracy Herp di ikki Eirci z Figure 15 Options window Image Filter tab Dnit Enable hapen Fia Emib Satuestion Fite 22 2 6 1 Parameters Panel The facilities provided within the Parameters panel allow the operator to set the conditions which filter out images for all enrollment verification and classification operations These parameters concern minimum head size head rotation limits and minimum tracking confidence values Also provided are filters for image quality concerning image sharpness and saturation 22 2 6 1 1 Minimum Size This text box sets the lower limit with respect to the size of the face that is used for enrollment verification and classification operations The size of the tracking box is established as a fraction of the height of the video display In the event that the tracked facial image is smaller than the specified size a verbal command is issued asking the individual to move closer to the camera The valid range for this
71. gins to behave erratically option settings may have been changed to an unserviceable combination Click the Restore Defaults button to restore all settings to their default installation values This button effects the settings defined within all tabs including the General tab 22 2 2 Enrollment Tab SS uu e The tabbed section entitled Enrollment allows the User to adjust parameters that control the enrollment operation Note that the option is provided to perform enrollment following a successful verification This allows the system to update the template continuously during normal usage acaLink Leva TIS FACIAL Figure 10 Options window Enrollment tab Descriptions of the various parameters that may be adjusted for enrollment operations are as follows wwen Help File 22 2 2 1 Parameters Panel LGWaa7GP The facilities provided within the Parameters panel allow the memm aj operator to set capture frame intervals thresholds for pattern i eo eet variance on image capture and enrollment timeouts 22 2 2 1 1 Frame Interval During an enrollment operation the system collects an image from every n video frame that is captured This helps to ensure a greater variation across the set of enrolled images This parameter sets the image frame interval for image capture 22 2 2 1 2 Image Difference Threshold Each image that is captured for enrollment must be sufficiently different from the previous en
72. gy the better Yes exclusion systems are capable of achieving the success rate necessary for those kinds of decisions For the most part biometrics appears to be a technology whose time has come from the marketing viewpoint It is suggested that the biometrics be used as a TOOL which is used to CONFIRM identity so not as the primary identification Business Week 2003 Why Biometrics Is No Magic Bullet Available online at http www businessweek com technology content jul2003 tc20030722 2846 tc125 htm Zephyr Analysis to determine the ideal biometric Hainlsoan a i rran P Yaslrr aran O O O ee Bee l ee Mj Fiai me fu New real time security alternatives are a reality today with the I CUBE SYSTEMS lighting fast Face Recognition System A leading developer of mission critical biometric solutions l Confidential Power Automation Printed 5 29 2005 MPRE NEN aT aT ee N HCUBE Page 43 of 104 Dawa Sewbalak CUBE SYSTEMS is committed to leadership responsiveness and unparalleled results We are driven to empower our clients and partners to go beyond tradition to create new benchmarks for security Accuracy is everything Facial recognition used for identification Biometric Intelligence Biometric security should be seen as an extension of human intelligence and not as a replacement for it because automated security will only be as good as the human intelligence that backs it up The danger of relying too heavil
73. hat captures a biometric sample from an individual Biometric Sample Raw data representing a biometric characteristic of an end user as captured by a biometric system for example the image of a fingerprint Capture The method of taking a biometric sample from the end user Comparison The process of comparing a biometric sample with a previously stored reference template or templates See also One To Many and One To One Claim of Identity When a biometric sample is submitted to a biometric system to verify a claimed identity Claimant A person submitting a biometric sample for verification or identification whilst claiming a legitimate or false identity Database Any storage of biometric templates and related end user information Even if only one biometric template or record is stored the database will simply be a database of one Generally speaking however a database will contain a number of biometric records End User A person who interacts with a biometric system to enrol or have his her identity checked Encryption The act of converting biometric data into a code so that is it unable to be read A key is used to decrypt decode the encrypted biometric data Enrolee A person who has a biometric reference template on file Enrolment The process of collecting biometric samples from a person subsequent preparation and storage of biometric reference templates Enrolment Time The time period a person m
74. ications ccecccccseeccceeeeeeceseeeeseeseeseeeeeeaaeeeeaeaees 36 14 LPR Equipment Quote cccccccccseececeeeeeceecetseeeesseeeeteneeeseaess 37 15 Software QUOTE epia erpa nesnesi EnA E A E EEEE 40 15 1 FACE RECOGNITION SYSTEMS aann0000annnnnnnooannnnnnnnnnnoonnnnnnnrnnerornnnnnnnrrnrrrnnnnne 42 16 Privacy dISCUSSION essorer nesens ann a ENE aAa ai 44 17 LICENSE PLATE RECOGNITION LINKED TO FACIAL VERIFICATION USER MANUAL ccccccecceeeseeeeeeeeeeeeeeeeeeeeenenees 46 18 Installation Quote ccccceeccceceeeeeeeeeeeeceeecetseeeeeeneeetseeeesaeeeeeees 56 19 Equipment Requirement cccecesecceeceeeeeeeeeecaeeseeeaeeseeeeeesass 57 20 Digital Recorder Integration cccccceccsecceeeceeeseeeseeeeeeeeeeeaeens 58 21 FACIAL IDENTIFICATION SOFTWARE USER MANUAL 59 21 1 Howto use l CUBE Facial Search cccceccccceeceeeseeeeeseeeeseeeeeeseeseesaeeeesseeeessaes 59 Bild ENO eE E EOE DEEE A E 59 Confidential Power Automation Printed 5 29 2005 Dawa Sewbalak akte BUCEO s decentenect aotcacienantancacdsciteheiacen setae oats tbaaaicaneinaue e EEEE EE 59 ek eT a E ee ee ne ee ee 61 21 1 4 Browse Database cccccccccsecccssceceeeeeceeeeceeeeceecessueessueeeseeeseeeeseeesseeesseneess 61 22 FACIAL VERIFICATION SOFTWARE USER MANUAL 63 ZA BaSe OPE 2 i 8 9 Cee eee ee eee eee ne enn eer E Meee 63 Ze Nel Gonnecting to Servel sssrin epei eea asado 63 2212 oS
75. idential Power Automation Printed 5 29 2005 TE Oe aT ne a CUBE Page 100 of 104 Dawa Sewbalak 25 Special Terms and Conditions This offer is subject to our standard conditions of agreement available on request Exchange Rate Rate used US 1 00 The prices quoted below are subject to exchange rate variations and will be based on 80 of the quoted price Any variation from this rate of exchange on the date payment is received by us within five 5 clear EDUCATIONAL FACILITYing days the date selection will be at the discretion of I CUBE is for the account of the purchaser The amount to be adjusted accordingly is 80 of the purchase price The rates of exchange which will be applied will be those quoted by EDUCATIONAL FACILITYers nominated by I CUBE Delivery Estimated 5 to 8 weeks from date of receipt of your payment Terms See Below Validity This offer is valid for 30 days from the above date VAT Prices EXCLUDE VAT General Terms and Conditions of Sale from I CUBE The Purchaser of products from I CUBE Integrated Intelligent Imaging herein called the Company is bound by the general terms and conditions below The General Terms and Conditions described herein together with the Company s price list and or proposal and or any incorporated documents shall be read together as constituting the sales contract IN THE EVENT OF ANY CONFLICT THESE 1 1 1 1 2 1 3 CONDITIONS OF SALE SHALL GOVERN LIMITE
76. ions of Agreement attached to the quotation If for any reason this agreement is not attached to the quotation a copy will be supplied on request Courier Delivered Approximately 5 to 8 weeks Not firm Exclusive of VAT import duties surcharges excise duties and any other ad valorem costs as specified in the Customs and Excise Act No 91 of 1964 and Amendments thereto are excluded Ordinary Customs Duty means any duty specified under Part 1 of Schedule 1 Import Surcharge means any duty leviable under Part 4 of Schedule 1 Ad Valorem Customs Duty means any duty specified under Part 2 Section B of Schedule 1 However if these ad valorem costs are quoted they should be considered only as a guideline of the costs ruling on the date of quotation If applicable these amounts will be invoiced and documentary evidence provided US 1 00 Any variation from this rate of exchange on the date payment is received by us within five 5 clear EDUCATIONAL FACILITYing days the date selection will be at the discretion of Protea Electronics Pty Limited is for the account of the purchaser The amount to be adjusted accordingly is 80 of the purchase price The rates of exchange which will be applied will be those quoted by the EDUCATIONAL FACILITYers nominated by I CUBE Delivery and export from country countries of origin of items requiring export licenses is subject to these being granted by the governments of the country countries of
77. is fine Or you may need mark the eyes manually if the system was unable to track the eyes 21 1 2 Batch Enroll You can do batch enroll to enroll multiple image files Confidential aga n y s Power Automation a A F f i a i ea m 3 Printed 5 29 2005 vl TERME aT e ee ee You have to prepare batch enroll text file The batch enroll file is a comma or TAB delimited text file It provides personal information and image file location For personal information you need have at least First Name and Last Name Image file location must be file name with full path So at least you need 3 fields For example Bob Smith c images jpeg 1 jpg John Brown c images jpeg 2 jpg You have to enter field ordinates based on your batch enroll file format According to previous example we have to put 1 after First Name 2 after Last Name 3 after Photo File and put O for all fields you dont use Click button to find the batch enroll file Click se ee er E Fi o a eo er Lar Harm pm o ie Biss fii f be _ qqa Emi a F Ho fae Trem Bo mooo ooo a Load to load the file You can preview the batch enroll file by lt and gt button By right you should be able to see the photo in that little white box If not check if the Photo File field is correct Note we need
78. ise thieves and prevent illegal entry Current system that allow abuse 12 5 Legislative EDUCATIONAL FACILITIES need to do everything in their power to prevent thieves from entering the EDUCATIONAL FACILITY EDUCATIONAL FACILITIES who do not take their public role seriously are quick to feel the backlash with serious consequences against those who are found to have a problem Implementing the proposed solution allows the PORT ACCESS CONTROL to show that they are doing everything in their power to prevent criminals con artists or terrorists from gaining access Please see the MBA dissertation Casino Exclusion Technique Exploration Framework Development by Barry T Dudley for detailed review of all exclusion techniques Send e mail to mba i cube co za to request a copy A PDF file of 2MB will be sent to your E Mail address Confidential Power Automation Printed 5 29 2005 MTF RFETE D A1 jH HCUBE Page 31 of 104 13 LPR Scope Of Work Dawa Sewbalak To supply and install a digital video surveillance and management system based on LPR and facial recognition that will provide authentication of the vehicle and video images of the entry and exit of all vehicles the driver vehicle as well as general surveillance of the access The LPR area The LPR system would replace security manually entering the data information would come from an image captured on the DVR via DDE All drivers facial images and vehicles would b
79. ject of public attitude It would be worthwhile study on attitudes and biases within the various segments of the biometric community for and against large scale biometric systems Some do not see it within their business interest for there to be rapid progress toward large systems since they may not feel that their technology or product is yet positioned to be competitive or dominant or are concerned that a niche they occupy or intend to occupy will be squeezed out by systems of more general application Cf Betamax vs VHS Mac OS vs DOS vs Windows etc The in depth study of the problems of privacy is beyond this study see Westin A 2001 for more information New technology is boosting biometric surveillance Grossman 2003 and privacy may vanish forever It is possible that legal and political issues such as privacy and data access could hinder the application of biometrics Lee 2003 Most of the public polls suggest that there is nowhere near the opposition to exclusion techniques that is Pe anaes claimed Very little effort has Wy ba ll been made by the i gt Cow bags government the press or the niim exclusion industry to explain TA eie irage and to distinguish exclusion E Corpae tw DY techniques from the controls that ought be placed on informational databases The result is that public concerns on the collection use and release of data are being largely ignored Privacy concerns are very difficult to addr
80. l button C eT First Name This enrollment operation is used to either add a new user or append new Last Name enrollment images to an existing user User ID sis This operation collects facial images from video input and uses these images Images fioo in the generation of the biometric template W Cancel Fill in the SAME information as you entered from the LPR system Enter Name ed First Hame NP17 27 Last Hame Bary ENSURE THIS INFORMATION IS UserID Dudep CORRECT H Images ico cancel When you click OK the person will be captured please ask them to be patient and they must be in their car looking at the camera If the wear glasses please ask them to have them both on and off please ask them to move and smile at the camera to ensure good images are captured Enter the User s first name into the First Name textbox and last name into the Last Name textbox Enter the User s ID into the User ID textbox Enter the number of facial images to be captured into the Images textbox The default value is 100 Click the OK pushbutton During the enrollment process facial images of the user are collected and added to the FRS database The progress of the enrollment operation in terms of number of facial images collected is displayed as a green progress bar within the Confidence Window located above the video display area Power Automation Printed 5 29 2005 ETF ee es Ap TPHA ed Se
81. l and hopefully exceed your expectations The License Plate Recognition linked to facial capture and verification solution allows Image trigger via a loop on the road Multiple image capture by a high resolution black and white camera License plate finding within the image License plate number reading at an accuracy over 99 95 with multiple cameras used in stereo Linking of the license plate with car colour and shape if required Linking of license plate number with a name and display of this information on ELECTRONIC DISPLAY if present when the vehicle is detected Linking of license plate number with a facial template allowing facial verification and automatic updating of the template as the person ages changes Log of number and image of all vehicles and driver who enter and exit the facility Image capture of the person from a high resolution colour camera Face finding within the image Searching central database and display of results eee Confidential Power Automation Printed 5 29 2005 OAT PE aT AT eo CD CUBE Page 9 of 104 Dawa Sewbalak Entry of all images and details that are not in the database Linking of the face with various fields such as ID no name Birth Sex Eye hair Colour Ht Wt Address City State Zip Country for entry into the suspicious persons database As leading specialists in using visual technology to solve customer problems and increase profitability we propose th
82. l commence upon tender of delivery to Purchaser Confidential Power Automation Printed 5 29 2005 2 1 2 2 2 3 2 4 ee ee 1p j THA k i 2 SOFTWARE LICENSE Page 101 of 104 Dawa Sewbalak GRANT OF SOFTWARE LICENSE The Company grants a limited non exclusive license to use not own one copy of the purchased software per unique computer under the Licensee s custody or control and subject to the following restrictions and conditions of this Agreement Sole Remedy The sole remedy of the Licensee for any damages related to use of the software shall be the replacement of the software or a refund of the value of the software at the Company s option provided that the Licensee notifies the Company in writing within one year of the purchase date Updates The Company will provide Software maintenance and or updates for a period of one year from the purchase date Thereafter should the Company elect to provide maintenance or updates the Company may charge a fee in an amount determined by the Company for such products or services or may waive said fee at its sole election TITLE TO SOFTWARE All title copyrights and trademarks in and to the Software including any accompanying printed material and any copies of the Software and all enhancements modifications and updates to the Software are owned by the Company 3 DELIVERY and SCHEDULE 3 1 3 2 Dates of shipping delivery or completion as may be stated in the C
83. l entitled New Registrant adjusts parameters tii 0 regarding the generation of biometric templates i e training Reigi io performed immediately after a user has been enrolled peace O These training parameters are also applied when re kii generating the biometric templates across all enrolled users initiated by clicking the Database Reset Database menu items The first text box sets the number of images used Carcel within each training cycle and the second text box adjusts the number of training cycles The third text box entitled Learning Rate sets the learning rate that is applied during the generation of templates The learning rate should be set within the range 5 to 40 The last text box sets an optimization level that is applied this value should also be set within the range 5 to 40 22 2 7 6 Continuous The lower panel entitled Continuous adjusts the same training parameters as those indicated above These settings are applied when performing the continuous training operation which updates biometric templates continuously using a background processing thread Confidential Power Automation Printed 5 29 2005 Page 81 of 104 Dawa Sewbalak 22 2 7 7 Image Parameters Prior to generation of the biometric template i e training the system will manipulate enrolled facial images in order to provide a more robust set of training images This process randomly offsets resizes and rotates the enrolled images The
84. landard VICW sspears pennn enr rn aE EEEE T NaS 64 22 13 Fal Screen VIEW eee iea A eee ee 68 22 2 Setting Operating ParaMeterS ccccceccccssseceesseeeeeeeeceeseeeeesseeeeseeeesseeeeesaeeeesees 69 eee ASIF TA aaecean state a E E 69 2222 Enrolment FAD rionn sede deosaucencusudeancceaceteneumanaseonce 70 ZA2 3 TORG WA res cits ara tletescutnanesa neactdauacuueouaesorhunnabyaanioaniehiasiaeat aoleninaeaenciie 71 2224 Classify TaD cen cearccctiqnndvcterseseecarchtcentitiecanancinaniceeacldencisaen aabhcisey ewes magtencetidtneh domsncanc 73 ae eo VEN aD eee eee ee ee eee eee ee eee 74 CO Made ICE WA cea E E A O 19 22 2 7 ES SIONS oisnean aroan E onnsa inr EE NTE anebo 76 Eaa e E A E A AE E E EE nous 80 223 OGNINA S E i E EER E 81 22 3 1 Editing the Database vee ete eee ee eee ne eee eee 81 Zs Pong Sgt eaceee os nace EE E 82 2239 Imponing US CNG seacrasannas casa nemenanceinincaconmbacinenpnceinanpaniemascinaneetioaadeuntananaubaidateaesennd 82 22 3 4 Regenerating Templates cccccccccsecesseeceeeeseeeceeeeeeecseeeaseeseueeaueecsueeasensaes 83 22 3 5 Pre Registration Of Users ccccccsecccsececsseeeceeeeseeeeceueeesaeeesseeeseusesseeesneneess 83 22 4 Enrolling from Static Images ccccccceeccceeeeeecceeceeeeeeceeseeeeeseeeeeseeeeesaeeeessaeeeesees 84 Cee ACIN LOO 16 WG asee E O E a 86 22 5 1 Showing the Activity Log cccccccsseeccseeeeeeeeeseeeeseeeeeeeeeeseeesseeeseeeeeseeeeseeeess 87 22 5 2 Filtering an E
85. le so that the biometric system can be more or less strict depending on the requirements of any given biometric application Throughput Rate The number of end users that a biometric system can process within a stated time interval Type I Error In statistics the rejection of the null hypothesis default assumption when it is true In a biometric system the usual default assumption is that the claimant is genuine in which case this error corresponds to a False Rejection Type II Error In statistics the acceptance of the null hypothesis default assumption when it is false In a biometric system the usual default assumption is that the claimant is genuine so this error corresponds to a False Acceptance User The client to any biometric vendor The user must be differentiated from the end user and is responsible for managing and implementing the biometric application rather than actually interacting with the biometric system Validation The process of demonstrating that the system under consideration meets in all respects the specification of that system Verification Verify The process of comparing a submitted biometric sample against the biometric reference template of a single enrolee whose identity 1s being claimed to determine whether it matches the enrolee s template Contrast with Identification WSQ Wavelet Transform Scalar Quantisation A compression algorithm used to reduce the size of reference Conf
86. lient application running locally These components provide the core biometric template generation data storage and communication to external Client applications For a more in depth description of please request the appendix Confidential Power Automation Printed 5 29 2005 OT Oe AT j THN ee CUBE Page 18 of 104 Dawa Sewbalak 8 FACIAL VERIFICATION Equipment Requirement The software is compatible with Microsoft Windows 2000 Professional and Microsoft Windows XP Professional The minimum system configuration requires a video capture card compatible with DirectX 8 0 in addition to the standard PC hardware Minimum hardware requirements are listed below I CUBE FACIAL VERIFICATION Client Microsoft Windows 2000 Professional Service Pack 4 or Microsoft Windows XP Professional 3 GHz Pentium 4 Processor 1 GB RAM I 1 TB HDD CD ROM Drive WDM _ compatible video capture device I CUBE FACIAL VERIFICATION Server Microsoft Windows 2000 Professional Service Pack 4 or Microsoft Windows XP Professional 3 GHz Pentium 4 Processor 1 GB RAM 1 TB HDD CD ROM Drive WDM compatible video capture device The performance of the l BPASSIEP e a TERS CUBE FACIAL VERIFICATION CLIENT PC is subject to at least an ISDN connection to the central server The I CUBE FACIAL VERIFICATION requires a skilled operator who has either been trained or has read and understood the Il CUBE FACIAL VERIFICATION
87. log is password protected in the same manner as the Options dialog 22 7 Setting Up Speech By default soeech synthesis generated during enrollment and classification operations is disabled Refer to section for details on how to activate speech Two modes of operation may be selected for speech generation these being either a randomly selected speech string or a speech string that incorporates the User s first and last name To enter a new speech string or modify existing speech strings open the SQL Server database called FRSDiscovery using a database editor such as SQL Server Enterprise Manager or DB Artisan Open the database table entitled Hello Table Confidential Power Automation Printed 5 29 2005 Page 89 of 104 Dawa Sewbalak To create a new speech string type RANDOM into a blank record under the Name field Following this type in the speech string that you wish to add under the Speech field Note that the phrase Hello First name Last Name will be added by default to all speech strings Randomly selected speech strings are produced following classification of Users who have not been assigned user specific speech strings To assign a speech string directed toward a specific user type first and last name into a blank record under the Name field Following this type in the speech string that you wish to add for that specific user under the Speech field Note that the phrase Hello First Last Name will
88. ly rotated within the specified bounds 22 4 1 4 5 Minimum Size The minimum size that is used for scanning facial images specified as a ratio of full screen height 22 5Activity Log Viewer The Activity Log Viewer displays all events that have occurred within the Discovery Client Server installation Included in the log report is the time and date in which the event occurred the Discovery client machine at which the event occurred the associated user name if any and type of event The Log also records an image from the client when the event Occurs The activity events that are recorded within the Log are as follows Enroll Failure Face Enroll Success Reenroll Verify Failure Face Verify Success Face Classified Face Unknown Confidential Power Automation Printed 5 29 2005 TFS EE 1p eT ee Se HCUBE Page 87 of 104 Dawa Sewbalak Multiple activity logs data may be viewed using the multiple document interface Each view may be filtered to list events by event date client computer User name and event type The Activity Log may also be saved or restored from a delimited text file 22 5 1 Showing the Activity Log In order to display the activity log perform the following operations e Click the Windows Start button Viewer from the Start menu e Click OK on the Display Log Viewer will open showing i ee a all activity events E a i Figure 28 Discovery Log Viewer window a le lfc lI 22 5 2 Filte
89. nail of the tracked face is placed in the upper right corner of the Confidence Window Two bars will display the actual confidence values The bar on the top will display the highest confidence value produced across all users that are currently enrolled A positive confidence value is shown in green and a negative confidence value in red The second bar displays the difference between the highest confidence value that is generated for the tracked individual and the second highest confidence value The classification threshold is displayed as a tick mark below the confidence display bar The generated confidence value must exceed this classification threshold while the bar is green in order for the individual to be classified i e identified The Confidence Window shown below displays results for up to two tracked individuals Figure 4 The Confidence Window detail lf a user is classified i e identified their name will be displayed on the left of the Confidence Window If a user is not classified within a timeout period then Unknown will be displayed The system can be set up to verbally greet a user on successful classification See the Setting Up Speech section The system can also automatically register and enroll a user if a pre set classification timeout period is exceeded See the Enrollment tab description in the Options Screen section for setting this timeout period Classify mode will continue until it is
90. nd device regardless of whether verification is successful or not The external Wiegand device is generally a legacy door access control system 22 2 7 2 3 2 Passthrough on Verify Only The ID number will only be passed through if the user has been verified 22 2 7 2 3 3 Passthrough all Registered Cards 22 2 7 2 3 4 The ID number is immediately passed through regardless of whether verification is successful or not however only registered ID numbers will be propagated 22 2 7 2 3 5 Never Passthrough The ID number will not be passed through regardless of whether verification is successful or not 22 2 7 3 PECOS Interface The PECOS Interface extension allows the Discovery System to send alarms to the PECOS system and to wireless Pocket PC devices that interface directly to the PECOS system Confidential Power Automation Printed 5 29 2005 x ETF ee es Ap TPHA ed Se Page 78 of 104 Dawa Sewbalak The PECOS protocol is a TCP based message that allows external systems to send alarms and alarm information to the PECOS system Each message is atomic and requires no reply from the PECOS system PaE Options Messages are 7 bit ASCII text and require no handshake Beal Sete W Enab There are two versions of the PECOS protocol Each version defines a READY message and a MATCH message Version 1 is Seren Addess extended to include Pocket PC se Ct lt C ti S support The dialog for configuring Poti i aii the
91. ned criteria established to evaluate the performance of the biometric system under test Physical Physiological Biometric A biometric which is characterised by a physical characteristic rather than a behavioural trait However behavioural elements may influence the biometric sample captured Population The set of end users for the application Recognition The preferred term is Identification Record The template and other information about the end user e g banned Response Time The time period for a biometric system to return a decision on identification or verification of a biometric sample Score The level of similarity from comparing a biometric sample against a previously stored template Template Reference Template Data which represents the biometric measurement of an enrolee used by a biometric system for comparison against subsequently submitted biometric samples Template Ageing The degree to which biometric data evolves and changes over time and the process by which templates account for this change Template Size The amount of computer memory taken up by the biometric data Third Party Test An objective test independent of a biometric vendor usually carried out entirely within a test laboratory in controlled environmental conditions Threshold Decision Threshold The acceptance or rejection of biometric data is dependent on the match score falling above or below the threshold The threshold is adjustab
92. neral performance characteristics pertaining to learning speed and accuracy Comparisons with comparisons to traditional neural networks Convergence Illustrates the learning convergence characteristics that occur when learning over multiple training exposures or epochs Generalization Concerns aspects concerning generalization and interpolation of the stimulus response mapping Neural Plasticity Describes the process of neural pruning and re growth and illustrates performance gained through the resultant optimization of input combinatorics 23 3 General Comparisons The two cell cerebellar model within HNeT is compared against a commercial system based on traditional genetic neural networks The genetic neural network used in this comparison permits up to 2 hidden layers and accommodates 256 cells per layer The primary feature of this type of neural network is the genetic based search used to find the optimal configuration i e number of cells hidden layers interconnections etc The holographic quantum neural approach HNeT does not require a search process and learns many orders of magnitude faster than traditional back propagation or genetic based neural networks The HNeT system is compatible with Oracle Visual C Access Sybase Foxpro Visual Basic 23 4 The Monte Carlo Test Accepted by many neural network experts as one of the more rigorous tests when it comes to evaluating artificial neural systems In a Mo
93. nfidential Power Automation Printed 5 29 2005 VTE PER ATP EHAN mea Dawe CUBE Page 104 of 104 Dawa Sewbalak Proposal for PORT for ADVANCED VEHICLE ACCESS CONTROL SOLUTIONS Incorporating LICENSE PLATE RECOGNITION DIGITAL RECORDING and Biometric Facial Identification VERIFICATION SOLUTIONS Compiled Supplied by FARA ph penera _ Dawa Sewbalak a E Power Automation j E Mail power auto intnet mu Sunday 29 May 2005 Slane In conjunction with l Cube MERCH RIWE sims pe La PR iB eee G Confidential Power Automation Printed 5 29 2005
94. nse learning neural cells associate or map one set of analog values i e the stimulus fields to an associated set of values i e the responses When the stimulus is distributed over a time span one has spatio temporal learning The mathematical basis for HNeT permits vast numbers of stimulus response patterns to be learned and superimposed enfolded onto a matrix comprised of complex scalars called the cell s cortical memory In fact the number of values used to store cortical memory is often no larger than the number of values contained within a single stimulus pattern The mechanism for holographic storage displays a capacity to achieve extremely high information densities due to the fact that large numbers of stimulus response memories can be enfolded onto the same set of scalars in other words computer RAM Confidential Power Automation Printed 5 29 2005 ee ee ns kH a Page 97 of 104 Dawa Sewbalak 24 Definitions Active Impostor Acceptance When an impostor submits a modified simulated or reproduced biometric sample intentionally attempting to relate it to another person who is an enrolee and he she is incorrectly identified or verified by a biometric system as being that enrolee Compare with Passive Impostor Acceptance Algorithm A sequence of instructions that tell a biometric system how to solve a particular problem An algorithm will have a finite number of steps and is typically used by the biometric engin
95. nte Carlo evaluation the stimulus response patterns are comprised of random numbers The comparisons below use 5 input variables for the stimulus and one response variable with values uniformly distributed between 0 0 and 10 0 The learning convergence characteristics are shown for densities of 100 500 and 1000 stimulus response patterns respectively At these low pattern storage densities non linear capabilities of traditional back propagation and genetic neural networks are pushed beyond their limit Applying this standard test method one may evaluate three aspects of operation The first aspect concerns the stimulus resoonse memory capacity of the system the second concerns the recall accuracy of the trained cell and the third concerns learning speed All three performance figures are shown for a 160 MHz Pentium Il Confidential Power Automation Printed 5 29 2005 Dawa Sewbalak 2 0 5 ONE HNeT CORTICAL CELL ii 25 WITH GRANULE CELL PREPROCESSING az E EPPEN EOE 3 2 0 h T T 1 5 i i i ti q i E 1 0 i i 5 is ii a S training time 2405 seconds training time 1 28 seconds 23 5 Comparison 1 Learning 100 Stimulus Response Patterns After the initial genetic search training time applied to the genetic neural network is 40 minutes By comparison training time for the HNeT system is 1 28 seconds At a storage density of 100 patterns the HNeT granule cortical cell struct
96. o automatically gain entry The software will allow the guard to manually enter the number of people in a visitor vehicle plus link any existing card based access control system to visitor cars if required On exit the guard will then be prompted to confirm that the same number of people exit as came in The on site digital recorder which will act as a base station from which recorded images from all the cameras can be reviewed from remote locations With the POWER AUTOMATION systems Facial Identification software all visitors are logged allowing any suspicious person to be immediately searched against the suspects database and the homeowner s database or the known workers database immediately identifying the person Confidential Power Automation Printed 5 29 2005 r ee ee LOSE ORs Page sof 104 Dawa Sewbalak CLIENT Sunday 29 May 2005 i ADDRESS Att Mr CLIENT TITLE Dear CLIENT PORT ACCESS CONTROL DIGITAL VIDEO MANAGEMENT SYSTEM PROPOSAL Thank you for affording Decalink Ltda the opportunity to quote on the installation of a digital video management system at the ESTATE including the use of a Vehicle Number Plate Recognition System as detailed herein As leading specialists in integrated intelligent imaging we propose the attached solution which if implemented correctly has the potential to solve your customers problems enhance some of your existing security services unlock additional potentia
97. ocate a face in any of the JPEG images If the system is not able to locate a face or the assigned ID number is currently used it will not perform the enrollment operation for that image file Following batch enrollment the system will report the number of users enrolled and list the images that could not be enrolled within the file EnrollStaticlmage log Confidential Power Automation Printed 5 29 2005 TL rE ETF AFETE E 1p j dP a a HCUBE Page 86 of 104 Dawa Sewbalak x 22 4 1 4 Options Image preprocess options for the enrollment of static a m i ia e k images may be modified by clicking on the Options One x E Raaienidegest in button located above the main display panel This o will activate the following form Mineram Siew fo 135 Figure 27 Static Image Preprocess dialog OK Canca The form shown above displays the recommended settings and each setting is described as follows 22 4 1 4 1_ Image Count Multiple images are generated from the segmented face These generated images are randomly rotated resized and offset in terms of vertical and horizontal placement using the settings provided on this form 22 4 1 4 2 Offset Establishes the degree of vertical and horizontal offset expressed as a percentage of the segmented facial region 22 4 1 4 3 Resize Establishes the resizing range expressed as a percentage of the segmented facial region 22 4 1 4 4 Rotate degree The facial image is random
98. of an existing user Files used to add images to existing users do not require the file name format indicated above To enroll a static image for an existing user Select From File Selection Select the JPEG file from the file list in the upper left corner of the dialog Select the target users name in the user list box The JPEG image will appear in the display screen located on the right side of the dialog The users face will be tracked automatically If no tracking box appears around the face click the Retrack Face button e When the face is located in the image click the Start Enroll button 22 4 1 3 Batch Enrolling of Static Images Multiple users may be enrolled in batch mode from a set of JPEG images Each image that is enrolled will create a new user within the system The system will automatically assign each new users name and card number based on the filename of each JPEG file The JPEG file names must have the following format LASTNAME _ FIRSTNAME _ CARDID Jjpg For instance a JPEG file with the name Doe John _ 25463 will create a user named John Doe with card number 25463 To enroll multiple users from static images in batch mode e Select From File Folder e Select the folder containing the JPEG files from the file list in the upper left corner of the dialog e Click the Start Enroll button to initiate enrollment of all users stored within the file folder e The system will report if it was unable to l
99. of the system to the point of competency is provided at no additional cost Thanking you in anticipation of the quotation being accepted and your instructions to proceed with the compilation of the rental documents and thereafter installation of the system Please be assured of my best attention at all times Dawa Sewbalak Power Automation E Mail power auto intnet mu wiviw t Cube co za Integrated Inteligent Imagine el ee lg ahi TAE er Be ed rae B Fi z y a ide TEON Be pE Ei pi A E a Ties oe ea ee Mo LS ig BN to i E A a 2 ee lg Pa ieee roa Ba ao ea cox 1 SA St eg fest Se eel ea Sets eg LMS ad beac tas a it e m PSG ot Lf gi ea oe ETI T ar ARAT w ET eg 1g 3 RRESERSPERESLURAEAZHPEERSELE i R va ot g ea PE Thal Confidential Power Automation Printed 5 29 2005 Page 12 of 104 Dawa Sewbalak 5 Description of Requirement The requirement as stated is for access control to an PORT ACCESS CONTROL with MULTIPLE X gates Each gate is assumed to have one lane IN and one gate OUT possibly separate lanes E 1 lane IN 1 lane OUT or same lane bi directional e Where two types of vehicles drivers are present those in the database enrolled and those NOT present in the database visitors e Where both license plate and driver face are enrolled in the database access will be allowed based on verification of the driver s face with the license plate on ENTRY
100. ollows Fara ori 22 2 4 1 1 Threshold tl The output value recognition confidence eee lS generated by biometric templates range from Faure Lag 3 1 to 1 with 1 indicating high confidence Classy Tiaoa pec i in recognition This threshold establishes the set point above which an individual is Gormnal positively identified i e classified This threshold is applied in conjunction with the difference threshold described below 22 2 4 1 2 Difference Threshold This threshold pertains to the difference between the highest confidence value that is generated among all individuals that are enrolled within the system and the second highest confidence value This difference must be greater than the specified threshold in order for the system to register a positive identification 22 2 4 1 3 Frame Lag The recognition confidence values that are generated across all biometric templates are lagged by video frame count This value establishes the degree of digital lag that is applied to the recognition confidence values when the system is operating in classify mode 22 2 4 1 4 Classification Timeout This value specifies a timeout condition that is applied to the classification operation In the event that the user is not recognized within the specified time period the system will either perform an automated enrollment operation or produce synthesized speech from a pre set Confidential Power Automation Printed 5 29
101. olographic principles HNeT cells provide both real time learning and dramatic improvements in performance over structurally more complex back propagation genetic neural networks Holographic quantum neural technology provides an exceptionally high connection per second or CPS rating in excess of 40 Million CPS on Pentium Ill processors This allows an HNeT cell assembly to learn and respond to several thousand input patterns in under a second The SL Platform a non programmers interface provides for training and designing supervised feed forward cell assemblies cells from ASCII or binary files The following provides a general specification list for the HNeT2000 Application Development System 23 2 Performance Features The following details some of the performance features that are unique to the HNeT technology The most basic cell assembly based on the cerebellar model is comprised of two synthetic neuron cells granule and Purkinje The performance aspects discussed are Confidential Power Automation Printed 5 29 2005 TI Me Lee Ll ee ee LCUBE Page 94 of 104 Dawa Sewbalak also characteristic of larger and more elaborate cell assembly structures within HNeT these more advanced structures providing further extensions to the core operation i e neo cortical model temporally based learning and unsupervised hyperincursive models A brief summary of the following performance features are covered General Provides ge
102. ompany s proposals are approximate and assume prompt receipt of all necessary information and reasonable cooperation from purchaser Delivery schedules are set from the date of receipt of system down payments The company shall not be liable for delay in its performance of the contract due to force majeure or causes beyond its reasonable control In the event of any such delay date of delivery shall be extended for a period of time equal to that lost by reason of the delay 4 COMPANY DESIGNS and STANDARDS 4 1 4 2 Because the Company is constantly improving its products the designs dimensions and weights shown in its proposals while sufficiently accurate for most purposes are subject to variation If extreme accuracy is required additional information and certification will be provided upon request after receipt of order The goods sold hereunder shall be manufactured to the applicable standards if any stated in the proposal documents In the absence of definite descriptive design criteria the Company s standards shall be applicable Confidential Power Automation Printed 5 29 2005 VALIDITY AGREEMENT SHIPPING METHOD DELIVERY BASIS DELIVERY TIME QUOTATION BASIS RATE OF EXCHANGE EXPORT LICENCES WARRANTY PAYMENT TERMS E amp O E Confidential Page 102 of 104 Basic Terms amp Conditions Dawa Sewbalak 90 DAYS Valid until 2005 08 19 This quotation is subject to the Standard Condit
103. on to the FACIAL VERIFICATION system which will check the database via DDE Checks the database for vehicles in the BLACK list which will cause an alarm Stops when fully at card reader if installed Person swipes and IF EXISTING 40 msec for a CARD confirms face via verification If single image not enrolled will enrol taking full capture frontal facial images and another image of the VEHICLE COLOUR Drives on Access card issued ticket now links facial images car colour license plate and date and time and lane 10 Goes to Idle a 0 5 second from the car entry a ees sefing setting 11 Next Vehicle triggers the sensor The timing of the system per lane is described below EXIT ae aa E The VEHICLE The system Vehicle enters EXIT LANE and approaches the sensor __ In idle waiting for the sensor activation a Passes over loop and Activates the sensor Nortec ee detector see documentation i capture Changes the illumination Captures 50 msec per image one or more images into memory capture Repeats the illumination change So Cals DLL on one or more images Identifies the car 80 msec per image identification 7 Logs displays transmits the license plate recognition to the FACIAL VERIFICATION system which will confirm this via the pre enrolled template via DDE Checks the LOCAL database for vehicles in the BLACK list which will cause an alarm and NOT allow boom to open Stops when
104. onnect window Ca e Enter the name of the computer hosting the desired Discovery Server application in the Server Name box e Click OK After the Discovery Client is run for the first time it will not require the Server name again and will connect to the Server automatically See the Options section for an explanation of how to connect the Discovery Client to a different Server Confidential Power Automation Printed 5 29 2005 MTE PFE r T Tp ee ee ca HCUBE Page 64 of 104 Dawa Sewbalak 22 1 2 Standard View The main screen of the I CUBE Discovery is shown below in Standard view The Control Buttons on the left side of the screen 7 Tey allow the Operator to run the program in the ecaLink Lemar rca various modes i e track only classify verify as igir well as adjust operational settings and change views The Confidence Window in the upper right of the screen provides a display of recognition confidence levels for Classify and Verify operations This window also functions as a progress bar during enrollment Add operations Tracking Boxes and Closest Matches are shown in their respective modes NUMBER PLATE seri ak hae Ee Figure 3 Discovery Standard view All of the control provided through the buttons are duplicated in the Control menu with the exception of the Full Screen button Full screen control is provided in the View menu Functions provided by the control buttons are described belo
105. ons window The Options window allows the user to modify operational settings within the Discovery system See the Options section for a detailed description of the Discovery system options The I CUBE Discovery may be setup to require a password in order to change options If the password option is set the following window will appear when PSP clicking the Options button Enter the Options password and click OK Figure 7 Password dialog The Options password protection is enabled and disabled in the General tab of the Options window Confidential Power Automation Printed 5 29 2005 Pod ie ite ee CUBE Page 68 of 104 Dawa Sewbalak 22 1 2 7 Full Screen View Click the Full Screen button or select Full Screen from the View menu to switch Discovery to Full Screen view n m 5 In Full Screen view the video window expands to fill the entire display All controls shown in Standard view will be hidden A subset of the controls are available in Full Screen view through a right click popup menu See the Full Screen View section later in this chapter for details 22 1 2 8 Exit The Exit button is located at the bottom right of the Discovery application when in Standard view This button closes the application Note that closing the Client application on the Server machine does not stop the Server application For information on terminating the Server Application refer to Chapter 4 EL esy Miery 22 1 3 Full
106. origin See Limited Warranty from I CUBE above The seller warrants products against defective material and or poor workmanship Prepayment See payment terms from I CUBE above Power Automation Printed 5 29 2005 ee ee ns kH a Page 103 of 104 Dawa Sewbalak 26 INDEMNITY The person or Company listed above agrees to indemnify hold harmless and defend I CUBE and its officers employees agents and representatives from and against Any liability loss and expense arising by reason of claims by government provisional municipal local or other authorities including Suppliers of equipment or any failure of those listed to comply with any Act of Parliament law ordinance regulation or bye law made with lawful authority by a government provincial municipal local or other authority provided that compliance by those listed with the above is required under the provisions of this Document at law or otherwise including without limitation failure of those listed to pay taxes duties or fees and Any claim liability loss or expense arising from actual or asserted infringement or improper appropriation or use by those listed of trade secrets proprietary information know how copyright rights both statutory and non statutory or patented or unpatented inventions or actual or alleged unauthorised imitation of the WORK of others arising out of the use or sale of materials equipment methods processes designs information or o
107. ort Valid port numbers range from 1 to 49151 22 2 7 3 3 Pocket PC Application Settings Confidential Power Automation Printed 5 29 2005 ET ET Sa x Ee es Ap AEA N errs Page 79 of 104 22 2 7 3 3 1 Send Alarms to Pocket PC Dawa Sewbalak When this flag is set and Version 1 has been selected and the extension is enabled the extension will send messages to the PECOS Pocket PC system on classification 22 2 7 3 3 2 Application Folder Path to the Pocket PC application on the Jakarta server that hosts the PECOS web application This value can only be set if the Send Alarms to PDA flag is checked The path is chosen by clicking the browse button to the right of the field 22 2 7 3 3 3 Alarm Folder Alarm sub folder of the PDA Application Folder for the PECOS web application This folder will receive all XML files generated by the PECOS extension This value cannot be set until the Application Folder is set The folder is chosen by clicking the browse button to the right of the field 22 2 7 3 3 4 Image Folder Image sub folder of the Alarm Folder for the PECOS web application This folder will receive all JPEG files generated by the PECOS extension This value cannot be set until Alarm Folder is selected This folder is selected by clicking the browse button to the right of the field 22 2 7 4 NetAlert Server This extension enables a Pocket PC equipped with wireless communication to receive HTML alarm messages and JPE
108. parameter is 0 05 to 1 0 In the figure above only faces larger than 1 10 the screen height are processed during enrollment verification or classification operations 22 2 6 1 2 Head Rotation degrees Establishes the permitted range in head rotation for facial images that are used for enrollment verification and classification operations In the event that the head is rotated away from the camera by an angle greater than that specified those images will not be used in any of the above operations In the figure above only images with head orientations less than 60 degrees from full frontal are processed during enrollment verification or classification operations 22 2 6 1 3 Tracking Confidence While the system is performing tracking a confidence value is generated in terms of how closely that image resembles a human face Tracked images that generate a confidence less Confidential Power Automation Printed 5 29 2005 JT ETF AFETE E Ap TPA k Se errs Page 76 of 104 Dawa Sewbalak than this setting will not be processed during enrollment verification or classification operations 22 2 6 1 4 Sharpness Filter Poorly focused images are rejected by checking this box 22 2 6 1 5 Saturation Filter Facial images that are either overexposed or underexposed are rejected by checking this box 22 2 7 Extensions Epean Genma Tracking imags Filler Extensions are plug ins to the system that provide extra Vey Claenl
109. phs are available on line and generated automatically thus simplifying the analysis of the operation of the system SeeService is an additional utility that keeps an eye on the recognition units SeeLane or SeeGate application In case it stops to respond rare cases it resets the application and restarts The utility also checks for updates in the Server and if there is a new file revision it automatically updates the application To simplify the installation a calibration tool SeeCal is available as a very useful tool The tool shows the camera video displays brightness contrast graphs and enables manual control over the I O card controlled illumination IR being used at night to highlight the PLATE 13 2Sample sequence of operation The timing of the system per lane is described below ENTRY TheVEHICLE iC Thessystem i Time Vehicle enters ENTRY LANE and approaches the sensor __ Inidle waiting for the sensor activation o O Passes over loop and Activates the sensor aa e Nortec loop detector see documentation i memory image capture 4 Changes the illumination Captures 50 msec per one or more images into memory image capture Repeats the illumination change ST O as Calls DLL on one or more images 6p Identifies the car Confidential Power Automation Printed 5 29 2005 ee ee es k a LCUBE Page 34 of 104 Dawa Sewbalak image identification Logs displays transmits the Identificati
110. pplies the power of digital holography within synthetic neuron cells Assemblies comprised of such cells have one to one correspondence with the primary cell structures of the brain These biomimetic structures provide the capability for truly real time learning and present a vast increase in stimulus response memory storage capacity To provide a practical example a cell assembly can locate and track human faces in real time A cell assembly can learn facial images in real time building within its memory all observed forms of an individual and subsequently identify that individual within a crowd even determine facial expression such as smiling or frowning etc This application is at the upper limit of technological capability when employing traditional methods Application of the basic two cell cerebellar model reduces the above task to a rather straight forward procedure The HNeT technology is not limited to face tracking identification but may be similarly applied to numerous areas within the medical sector process control automation defence financial etc NEURAL PLASTICITY NFA g SPATIO TEMPORAL LEARNING 23 1 HNeT Tools The HNeT system allows our developers to construct neuron cell assemblies and integrate these neural assemblies into applications The core of the HNeT system is a Dynamic Link Library DLL containing over 90 functions for creation of cell assemblies and customization of cells Employing h
111. previous section The options form is divided into eight tabbed sections 22 2 1 General Tab The tab shown below allows the operator to adjust options that pertain to continuous re generation training of the biometric templates used during classify and verify operations options for automatic brightness compensation enabling speech synthesis password protection on program variables changing the Discovery Server and resetting parameters to default values Figure 9 Options window General tab 22 2 1 1 Continuous Training Panel The Continuous Training option is available on the Discovery Server application only When this option is enabled re generation of all biometric templates within the Discovery Server database is performed continuously in order to improve the speed and accuracy of verification and classification operations Continuous training is performed by the FRS Training Server while it is running By default the continuous training operation starts automatically on startup of the Server To stop continuous training click the Stop button Note that training can be controlled directly through the FRS Training Server for further details refer to Chapter 4 22 2 1 2 Brightness Control Panel Brightness control adjusts the video capture driver to compensate for mean pixel intensity located within the region of the tracked face Hysterysis is applied to brightness control and limits the amount of compensation applied 2
112. r camera Face finding within the image Searching central database and display of results Entry of all images and details that are not in the database Linking of the face with various fields such as ID no name Birth Sex Eye hair Colour Ht Wt Address City State Zip Country Facial verification based on license plate or access card 7 2 Face Recognition System Description The Face Recognition System is used to assist in identification The surveillance operator manually compares the live or recorded images and the saved facial images against a database of previously saved face images with an operator reviewing the results and making the decision This means that the accuracy of the system is NO LONGER crucial as the system presents information to a human operator to make the final decision One is using the face recognition system to check if the person has been seen before with the operator looking at the results to check the match This is due to the fact that the following affect the results of ANY face recognition system Lighting quality of original image in the database temporal affects time and ageing glasses hats shadows hair style or lack of hair background size of face in the image emmmmm mm mmm and a WIDE RANGE of other environmental conditions One is using face recognition to assist in IDENTIFICATION of repeat trouble makers illegal entry shop lifters bad check passes and then
113. rd Skip Blank images means dont display records without image Confidential a jpeg file by clicking Save button image will be save as Power Automation Printed 5 29 2005 Page 63 of 104 Dawa Sewbalak 22 FACIAL VERIFICATION SOFTWARE USER MANUAL This chapter describes the basic operation of the I CUBE Discovery system and how to adjust operational parameters Most of the features described in this chapter are common to both Discovery Client and Server side applications Features that are available on the Server only are noted 22 1 Basic Operation The Discovery system for both standalone and multi Client installations provides tracking verification and classification functionality It operates in two views Standard and Full screen In Standard view all functionality is available through menus and buttons and classification values are graphically displayed In Full screen view the live video display fills the entire screen Program controls are available through right click menus 22 1 1 Connecting to Server Before running any of the Clients you must have the Discovery application running on the Server When a Discovery Client application is started for the first x time the User is presented with the Connect window A J i Discovery Client may be connected to any Discovery lt Server application i e database currently running on the Serves Harra JERE network Figure 2 The C
114. res cccccecccsccceeccceecceeccceeeceesceseeseeseeeeecueeceeeeeeeesseseeeesseees 93 23 93 General GOMPANSONS scastedevcdscaedesdanveceecotecnsnecessexededceeee per aa iaaea ri aaeei 94 234 The Monte Carlo Test cise ieccdcvnsascoudsedumnted ieee cesandecotsienerncnecesetedeutaotesentiemasiivecnne 94 23 5 Comparison 1 Learning 100 Stimulus Response Patterns cccccceeeeeeeeees 95 23 6 Comparison 2 Learning 500 Stimulus Response Patterns cccccseeeeeeeeees 95 Zoe TAC BIOO si see decd annie tecentaa A E E 95 24 DePINITIONS o oo ccc cc ccc ec eceeeuaeeeeeseueueueeeauaueueueavaueueueavaveueeuavavauaes 97 25 Special Terms and Conditions cccccsecccceseeeceeeeeseeeeeeeeees 100 26 NOE NINIT Y comexpaseenzeecavesaueesnacsnteascnensipeseestucacs demeeeeseesenoeeeaiteaie lt 103 Confidential Power Automation Printed 5 29 2005 B WYP REE TER ATP CUBE Page 4 of 104 Dawa Sewbalak 1 Confidentiality Clause Sunday 29 May 2005 Due to the strategic importance of this work it would be appreciated if the contents remain confidential and not be circulated for a period of two 2 years Sincerely Dawa Sewbalak Power Automation E Mail power auto intnet mu PEOLG i jz i ET IHUN F i jijiji reat CEITETETENEELEFENEETETAATETATE ETT TETON SLED COT Cie miiti Piipa MERC TRI es See 2 or ae Confidential Power Automation Printed 5 29 2005 atau aiai secs f
115. reviously enrolled by the system The recognition confidence value of the comparison will be displayed in the Confidence Window If this confidence value exceeds the verification threshold the user will be verified and any corresponding actions that have been preset during program setup will be executed i e relay contact activation NetAlert alarm Wiegand signal pass through The Verification Threshold may be modified within the Options window When in Verify mode the tracking operation is enabled automatically Click on the Verify button to disable the verify operation facial tracking will remain active Click on the Track button to disable the tracking operation Extensions have been provided which combine the verify operation with a Keyscan Wiegand Serial Interface and a NCD Relay model R42 R410 or R810 The Wiegand Serial Interface can be connected to various Promixity or Wiegand card systems to provide ID card or keypad based functionality If a card reader or keypad system has been set up the system will automatically begin verification on either keypad entry or Card swipe lf a relay system has been set up the system can close or open an external circuit such as a door strike on verification or classification See the Extensions section for more details regarding peripheral devices such as relays and Wiegand devices 22 1 2 5 Add New User Click on the Add button or select Add from the Control menu to add a ne
116. ring an Existing 1 s View pi amp A Md To filter the events in an existing view T oS e Select Discovery Log Filter window The Discovery dip du des wey d l all bs select the view that you want to filter then select Filter from the Window menu or click the Filter button The following Activity Log xj Filter dialog will appear ILN Figure 29 Activity Log Filter dialog The ia aowa vow bo daph a tubae bo the acihiy iby User Hare ealegey of atip are dala e Select the filter criteria You may filter on a single User name Client Computer the Activity Category or range of dates Selecting multiple criteria will form a union of all selected filter criteria e Click OK The view will filter the events based on the selected criteria 22 5 3 Opening a New View To open a new window with a different filter select New from the File menu in the Log Viewer or click the New button The Activity Log Filter dialog will appear Select the filter criteria then click OK The new view will open in a new window 22 5 4 Saving the Activity Log The data from the current view can be exported to a text file for later viewing Note that images are not exported when saving the activity log To export the activity log perform the following operations Confidential Power Automation Printed 5 29 2005 Se ee Os ee Page 88 of 104 Dawa Sewbalak e Select the view that contains the data to be exported e Sel
117. rmation double click on the user name in the list box The following Update User dialog will appear UpdateUser Figure 23 Update User dialog In the Update User dialog you may change the first name last name and card number To delete a user right click on the users name in the list box and select Delete Multi select may also be applied when deleting users To delete an image select by clicking on the image and a blue boarder will appear Click the Remove button The image will be removed from the system and is no longer applied during template generation You may multi select images for removal by clicking on the first image you wish to remove then hold down the SHIFT key and click on the last image within the image set All the images between the first and last images will be selected Click on the Remove button to remove the selected images Click the Remove all button to remove all training images for the currently selected user 22 3 2 Exporting Users Users may be exported from the Discovery database for subsequent import into a separate Discovery installation To export a user or multiple users select the user s within the list box right click and select Export Users A common file dialog will appear Enter the name of the file that is to contain the exported data and click Save All data including training images will be exported to the file 22 3 3 Importing Users Users may be imported to the Discovery s
118. rmation to be added to the database both locally and remotely colour drivers face SeeLane software application package including integrated SeeCarDLL recognition engine Confidential Power Automation Printed 5 29 2005 OT PER iT p eT HCUBE Page 41 of 104 Dawa Sewbalak SeeLane Product user s license POWER AUTOMATION System Scope of Supply see product leaflet for description and performance SeeLane is a PC based multi lane LPR system designed for low speed traffic The SeeLane system includes both hardware and software and can accommodate either one camera per lane standard configuration or two cameras per lane stereo vision for added reliability and security Database Remote Database camerafillumination ann fprrt rrr er ae ml Seis n Figure See Lane configuration RE Bee jir U Sier E ieej E E E A T a ds E ee A emg aie T ps TO INCORPORATE ONE LOGGING SOFTWARE mi j Be SS i a Confidential Power Automation Printed 5 29 2005 PL bo a ee ee ns kH a HCUBE Page 42 of 104 Dawa Sewbalak 15 1FACE RECOGNITION SYSTEMS Biometric technology and specifically face recognition which can recognize people from a facial image is becoming cheaper and more powerful as technology improves Biometrics comes in many forms The idea is said to date back to ancient Egypt when records of distinguishing features and bodily measurements were used to make sur
119. rocess from here Dawa Sewbalak The standard process after due consideration and finalization of various components is to formalize the agreement to move forward with an order to commence the project This would take the form of either signed acknowledgement of acceptance of the proposal a letter indicating acceptance or a formal order number Standard Ts amp Cs do apply but these will be reviewed depending on what is accepted A formalized project kick off meeting would be arranged at the PORT ACCESS CONTROL where we would discuss in detail the solution proposed and determine exactly what will be deployed where and also understand the areas of responsibility for all parties involved The typical Project documentation will outline the following Detailed scope of work Finalization if the design Responsibilities of all parties Expectations Deliverables Timings and milestones Financial aspects Requirements for communication Assumptions concerns etc Exclusions Risks and how they will be addressed Personnel development training Commissioning and implementation schedules Standards that must be complied with amongst other things NS Once the plan is agreed and signed final equipment orders will be placed on the manufacturers and the project planning will be completed Equipment installation will commence in accordance with the determined schedule as well as with the receipt of the equipment
120. rollment image in order to ensure reasonable variation among the images collected during enrollment This value sets the variance pattern difference threshold which must be within the range 0 0 to 1 0 22 2 2 1 3 Track Timeout The enrollment operation will terminate automatically if the system is unable to track the user for a period of time in seconds specified by this setting Confidential Power Automation Printed 5 29 2005 Page 71 of 104 Dawa Sewbalak 22 2 2 2 Verification Panel The facilities provided within the Verification panel allow the operator to set parameters concerning automated re enrollment performed following a successful verification operation 22 2 2 2 1 Enable Enrollment During Verify When this flag is checked Discovery will enroll a preset number of facial images during each successful verify operation This allows the system to continually adapt and refine the biometric template over time 22 2 2 2 2 Maximum Images This value determines the maximum number of additional facial images that Discovery will enroll during each successful verification operation 22 2 2 2 3 Confidence Threshold Images collected for enrollment during verification must have a confidence value less than this threshold This ensures that only images containing a new perspective view of the user will be enrolled 22 2 2 3 Enroll During Classify lf this control is checked and a classification operation exceeds the timeout d
121. ry may be viewed by clicking the Training Log button This log displays the user ID for individuals that have been most recently trained the training method i e background vs continuous and Coriri Carat Cae ara Ciri Liv dirs Ta TE Coira imira gare Coira Coriri Coriri Coriri Ciri Lidia aTe Baza 207 29 PH WlOvenog 20717 PH H1102003 207 11 PH 11102003 20705 PH 110i 2 0658 PH Aaaa 2 0653 PH Aaaa 2 OGM PH ANG 2 G1 PH BAAG 0EM PH EAG 2 06 28 PH BAA 2 06 22 PH B11020003 20615 PH 11102003 20606 PH 1110200 2 0689 PH Aaaa 2 0a PH HAD 2 od PH Open Training Server Manager the time at which the biometric templates were trained i e Hide Training Server Manager re generated as follows ae xl Control of the Training Server is also available by right clicking on the FRS Training Manager tray icon A popup menu will appear with the option to open and hide the Training Server Manager Selecting Exit will close the window and stop the Training Server Confidential Power Automation Printed 5 29 2005 i KTERE TET Ap Ea pe CUBE Page 93 of 104 Dawa Sewbalak 23 Biometric Intelligence Overview Biomimetic Intelligence is the science of understanding and replicating the processing mechanisms and structure of the brain Traditional neural networks have little or no resemblance to actual neurological structures and more importantly have proven to be very limited in capability The HNeT technology however a
122. s individually or through multi selection These file are saved using a file name that is generated from the corresponding textual data displayed in the left panel of the window as follows DISCOVERY _ Time _ User Name _ Computer Name _ Activity To save the selected images to JPEG files perform a right click operation on a selected image and click the Save Image menu item VIEW HISTORY Access the LPR LOG 1 Overview This application is a new Client application for the SeeLane vehicle recognition system a recognition browser for past and current recognition results It displays the recognition results Vehicle and event data lane date time and allows the user to view the images of previous results Its inputs are a History results the recognition log file log txt created by SeeLaneClient application b DDE messages new events generated by SeeLane SeeTraffic SeeRoad LPR systems Its outputs are displayed history and images Power Automation Printed 5 29 2005 atone tia ce GUBE Confidential Page 55 of 104 Dawa Sewbalak It does not save any files and will discard the collected data when the F application is aborted 2 2 1 It can also be used offline on another PC if the images directory is set the same name as the original directories 2 2 2 It can also run on line on a Central server assuming that the SeeData tool is used to copy the images across the network 4 2 Vi
123. s would build visitor confidence loyalty and satisfaction lower marketing costs increase margins and provide an opportunity for brand extension Schrage 2003 The LPR techniques must be able to be used to increase the loyalty of the users of the EDUCATIONAL FACILITY Please see the MBA dissertation Casino Exclusion Technique Exploration Framework Development by Barry T Dudley for detailed views of this role player Send e mail to mba i E cube co za to request a copy A PDF file of 2MB will be sent to your E Mail address 12 3 Security SUT H GATE Example of Cape Access Control The concerns of security are often at odds to the rest of the role players Is Khayalethu Makhotyana in the figure below making the best utilisation of limited resources Currently security places a high emphasis on the reaction to events rather than being proactive We believe that the proposed system can be spark which could propel security to become a greater strategic ingredient Rather than reacting to outside forces the security department could lead the way in solving the problems Tradeoffs exist among product and process choice versus the longer term operating choices regarding quality efficiency schedule and adaptability Adam amp Ebert 2001 The first users of these LPR systems in South Africa have already successfully applied managed and maintain a solution thus earning the respect of all the role players Confidential Power
124. sion There is existing equipment that may be used in this solution offering but we have not considered this at this time This will be reviewed at the kick off session where all the existing cable and equipment will be assessed and integrated where possible There may well be some information that has been omitted as a detailed site inspection has not yet been conducted There are direct benefits to security legal matters relating to security breaches operations as well as an additional marketing tool to enhance value to owners Confidential Power Automation Printed 5 29 2005 Page 14 of 104 Dawa Sewbalak 7 Solution Proposed The proposed solution would enhance the image of the PORT ACCESS CONTROL and would also provide huge practical security access control enhancements to the day to day operation The offering to address the stated requirements will be tackled as a complete entity but can be phased or broken down into its components and implemented as separate items if required Some technical terms will be used and a glossary of terms is supplied as an Appendix to this document to assist with understanding should it be required 7 1 Solution Design A basic matter of understanding the difference between Identification and Verification is crucial for the development of appropriate solutions Identification is the assessing of the car or object presented for identification and attempting to derive a match from the r
125. such as the poor exchange rate It is therefore not recommended to pay cash for a hi tech system like this as the capital will be tied up in a technology that has little equity and will be superseded within a three year period Due to the rapidly advancing nature of the computer technology used in our systems it is our policy to Rent our own locally developed and manufactured systems that will continually keep our customers abreast of the latest developments at affordable Rand based rates This is achieved by phasing in minor software upgrades over the Rental contract period We will however offer an entire software and hardware upgrade after the initial three year period To achieve this POWER AUTOMATION will cancel the outstanding twenty four rentals on the old agreement upon acceptance and approval of a new five year rental agreement which will be kept at an acceptable and competitive rate Whereby the latest updated system will then be installed The choice of a 0 or 12 escalating Rental agreement remains your choice However as this is a balloon rental the end total is the same It therefore does not give any benefit to pay the higher non escalating rentals in the beginning due to the fact that we will be upgrading the system at the three year period Furthermore due to the fact that the goods remain the property of POWER AUTOMATION and do not become an asset to the customer there is no point in paying off the goods Unlike the traditional lease r
126. t more accurate result But you must be sure they are the same person You can change target data by clicking Select Targets button By default it will search everyone in database Click Search button to start search The result will be showed after search is done 21 1 4 Browse Database Confidential Power Automation Printed 5 29 2005 atone tia ce GUBE Page 62 of 104 Dawa Sewbalak You can do quite a lot of things in browse database dialog box You can show eye marks by check Eye Marks checkbox You can let system re track eyes for current image by clicking Relrack Eyes button You can manually mark the eyes by clicking ReMark Eyes button You can re track all people in database by clicking ReTrack All button Note you need do this extremely carefully Coz for big database it may take very long time to finish You can delete this record by clicking Delete button You can save this image to UserlD jpg 1 Hroms Dalansspe FELA WE You can click Find button to find the images satisfied with certain condition After that you can click All button to make it back to all dataset You can type UserlD and click Go to go to that record You can browse the database by clicking lt and gt You can also type number of images to go and go to that reco
127. tAlert will function 22 2 7 4 5 HTML Template Files The alarm messages are sent to the Pocket PC in HTML format Templates have been used to allow the user to change the appearance of the HIML page The entry templates are stored as UTF 8 text The default installation will place these files in folder c andface The following three templates are used to generate each HTML page and either the pre configured or modified templates must be assigned 22 2 7 4 6 HTML header template The HTML source code that precedes the alert information Any header or meta information normally goes here The filename for the header template is header txt 22 2 7 4 7 HTML item template Used to display the HIML code for each alert The filename for the item template is entry_templtate txt xj 22 2 7 4 8 HTML footer template Gennel a The HTML code that follows the alert items The filename for pent _ the footer template is footer txt mmn W Training o D Learn Aate 115 Training options are only available on the Discovery Server l p application They are accessed through the Training tab iaaah l within the Options screen These training parameters effect Conima e the generation of biometric templates performed after facial A _ images have been collected during an enrollment operation hanbe ol Cys 120 T Lawring Rata 15 Figure 21 Options window Training tab price Level Percer T5 22 2 7 5 New Registrant doran The top pane
128. te salsancsien lt incthnidahesccadmatecasiualed denmopsadstecastedinecsreastsuluadonactnse 19 16 Explanation Of BIOMEHIICS ccccccecceeecseceeeceeceeeseeeceeceeeeeeseeesenees 20 8 3 Normal process from here sie ttcectencessvensindidwonsndecdwwrswandendnoanadsbvaruliiensndedestuveoninezzeaes 21 9 EQUIPMENT Considerations cccccccceeseceeeeseeeeeeeeteeeetens 22 OPN NS asc nec sees neat sat sors eae lose cce eco enne arte nase hsecanneeedenescneeaseesenepioasenucecteouaneeteee 25 9 1 PICCCDUANGS ara ened aunties ucubnetsinatoenneyandes susGaeectcpeuachta E 25 9 2 Standard Terms and Conditions ccccccccssccccceeccceseeeeeseeeeecseeeeeseeeeeeseeeeesseeeeesees 25 10 PROM hell ODIO aane E E EA 26 10 1 BENEFITS OF POWER AUTOMATION RENTAL 0 0eeeeeeeeeeeeeeeeeeeeeeeeeees 26 11 CONC USIO ee E A 27 12 Introduction Role PlAayers cccccccseceseeeseeeeeeeeseeeseeeeeeeeseeees 29 lel WANG NO ee E EE E EEE 29 lee MAON e E E A E 29 2o SOU eenen EEE E AEE 29 eA OPON a A E E ee 30 leS CEO e E S E E 30 13 LPR SCOPE OFO eesriie E E E E 31 13 1 Full description of system topology offered cc ecccecceeceeeeseeeeeeseeeseeeseeeseeeneees 31 13 2 Sample sequence of OPEratiON cccceccccsseecesseeeceeeeeeeeeeeeseeseeeeeeeesseeeeeseeeesaaees 33 13 3 Suggested LPR PrOCOQUire axccsccstecscecendcesescecdenccceevedsceedehce cansdeveateaugebdeauentecseuessedes 39 13 4 Schedule Of Equipment Specif
129. text box and click the Add button The Training Server will be connected to the FRS Central Server and assigned training tasks Confidential Power Automation Printed 5 29 2005 ETF ee ee a aT ee Se Page 91 of 104 Dawa Sewbalak 22 9 4 2 Removing a Training Server To remove a Training Server from the FRS Central Server select the Training Server from the list of connected Training Servers and click the Remove button The Training Server will be removed and the training load will be shifted to the other Training Servers 22 9 4 3 Controlling Continuous Training The purpose of the Training Servers is to continuously update biometric templates using enrolled images To stop continuous training on a Training Server select the Training Server from the list of Training Servers and click the Stop button To start continuous training on a Training Server select the Training Server from the list of Training Servers and click the Start button 22 9 5 Selecting the FRS Database The FRS Central Server may be connected to any FRS database on the network The database provides biometric templates to the FRS Central Server for use in the clients and stores all enrolled facial images used in template generation The FRS Central Server can be connected to only one FRS database at atime Database Databare Server DEHD To connect to a database click the ellipsis button on the Database line The Database window will appear Dala
130. ther things including construction facilities furnished those listed or its nominated personal in or for performance of the WORK and Any claim demand cause of action loss expense or liability on account of injury to or death of persons including the employees of the I CUBE or damage to or loss of property including the property of the OWNER arising directly or indirectly out of the acts or omissions to those listed or its SUB Contractor s or the employees or any thereof in the performance of the work including without limitation such claims loss of liability arising from the use or operation by those listed of construction equipment tools scaffolding or facilities furnished to those listed by I CUBE to perform the work irrespective of whether party to be indemnified was concurrently negligent whether actively or passively and including any expenses and attorney s fees incurred by l CUBE for legal action to enforce those listed indemnification obligations under this clause but excepting where the injury or death of persons or damage to or loss of property was caused by the sole negligence or wilful misconduct of the party to be indemnified and Any claim demand cause of action loss expense or liability on account of actual or alleged contamination pollution or public or private nuisance arising directly or indirectly out of the acts or omissions to act of those listed or its SUBCONTRACTORS in the performance of the WORK Co
131. tial Power Automation Printed 5 29 2005 KTF AEE TEE A1 aT gt ee Se fatty Page 96 of 104 Dawa Sewbalak within sections of the brain referred to as the neo cortex gray matter or outer layer the cerebellum near the base of the brain and the hippocampus The HNeT system allows one to construct cell assemblies ranging in capability from supervised feed forward systems to more advanced spatio temporal and hyperincursive models CORPUS CALLOSUM CEREBRAL CORTEX _t FRONTAL LOBE E THALAMUS THALAMUS gem MEDULA OBLONGATA CEREBELLUM f 7 4 SPINAL CORD MUSCLE HNeT cells have been given biological names due to their similarity to specific classes of neuron cells i e the granule stellate Martinotti pyramidal and Purkinje cells This section is provided for a more technically inclined audience Although the mathematical basis for HNeT is somewhat abstract one does not require an in depth understanding of the theory in order to design and build applications using the HNeT2000 Application Development System It is important that one understands how stimulus response information is presented to the system and how the various types of holographic quantum neural cells interact with each other A stimulus response pattern or memory may be represented by a set of values reflecting conditions or states measured within an external environment such as pressure temperature brightness etc During stimulus respo
132. uantifiable return on investment Each phase in the installation is justified by its own standalone ROI Moreover look closely at the effect the shift to the proposed infrastructure will have on near and long term infrastructure efficiency image and customer satisfaction sa Ppp tee arp Appendices Confidential Power Automation Printed 5 29 2005 Ue Ud ee CUBE Page 29 of 104 Dawa Sewbalak 12 Introduction Role Players 12 1Introduction The role players involved with the selection and implementation of an integrated LPR system which can involve security EDUCATIONAL association marketing and CCTV surveillance systems need to involve all departments However the role players affected have a variety of criteria they might apply to the selection of a solution Within the PORT ACCESS CONTROL there are at least four primary areas that would be affected by the introduction of facial recognition monitoring techniques These are e Marketing improving the experience of the visitors by recognising important guests e Security ensuring the safety of people and property e Operations LPR systems have a crucial role by speeding up identification and e Legislative identification of any threat to the EDUCATIONAL FACILITIES by thieves terrorists or con artists 12 2 Marketing The marketing department could use these n proposed solutions to enhance the brand building experience to the bonel of the EDUCATIONAL FACILITY Thi
133. umber and name the imported image set will be merged with the image set currently stored within the database 22 3 3 2 Import Log The progress of the import is saved to a text file Any rejections will PESA Aa appear in the import log The import log resides on the server and Penonal into stored as importuserslog log Frat Mame Pulp 22 3 4 Regenerating Templates os Select Retrain All from the File menu of the Edit Database window ee in order to clear all biometric templates and regenerate i e train Coad ER the templates across all users currently stored within the Discovery database Once a retraining operation has begun the biometric poet templates across all users are cleared and a classification or verification operation will not function until all templates have been regenerated Regeneration of templates requires approximately 10 Pre regestraticn ream ie seconds per user Sect ie Pompeys iat 22 3 5 Pre Registration of Users Users may be added registered within the Discovery database without performing an enrollment operation i e facial image io collection This operation is referred to as pre registration Pre registered users are enrolled at a later time through use of either the Wiegand interface i e card reader or keypad device or the Add push button control located on the main screen To pre register a user or group of users select Pre register User from the Database menu The following Pre
134. uration default 6 seconds the system will present the Enter Name dialog and enroll a new user For information on classification timeout refer to the section describing the classify options If this box is not checked users will not be enrolled following a classify timeout condition and the system will soeak a pre set phrase i e Could you please register with the receptionist thank you 22 2 2 4 Enroll Pre Registered Only If this box is checked the system will automatically enroll a user who has been pre registered the first time they perform a verification operation via the Wiegand interface While this option is enabled the Add button is removed miy Cera Erpen 22 2 3 Tracking Tab Bee Tig image Fim Erien FEH The tabbed section entitled Tracking allows the operator Fokan T pamati E to adjust parameters concerning facial tracking These bir Tia O parameters apply to all modes of operation which include tole i Tracking Enrollment Classify and Verify The jo tracking subsystem is based upon a Knowbot structure a o ome 5 consisting of a geometrical arrangement of retinal viewing ta a oe areas This knowbot structure performs a scan over the video capture frame looking for peaks in the recognition topology The neural assembly forming the brains behind the retina has been trained to recognize human faces The system also uses basic cues to estimate where a face may be located These
135. ure is 100 times more accurate and 2000 times faster than the traditional neural network 23 6 Comparison 2 Learning 500 Stimulus Response Patterns Increasing the 3 0 number of Mogna as stimulus response amp aman ee ag patterns causes i GENETIC NEURAL NETWORK ONE HNeT CORTICAL CELL a i i 1 5 apean ort the genetic neural network to 3 an E approach a state 3 6 j X s a of saturation At ab Len training time 139010 seconds this level of storage density traditional neural networks break down Learning capacity of the HNeT granule cortical cell combination is unaffected by the increase in storage and displays a convergence similar to the test involving 100 patterns 23 6 1 1 Comparison 3 Learning 1000 Stimulus Response Patterns At 1000 stimulus response patterns 5 na the genetic neural 5 2b network is unable 2 20 i GENETIC NEURAL NETWORK f ONE HNeT CORTICAL CELL to achieve any 2 o as measurable level ww l of convergence even after 20 3 ne ee hours of training training time 72000 seconds ms training time 15 seconds The rapid learning characteristic of the HNeT system is again unaffected by this increase in storage density 23 7 The Biology The following provides an overview of HNeT biomimetic intelligence Biomimetic intelligence models cell inter connectivity and signal processing aspects of actual neuron cell assemblies Confiden
136. user manual 8 1 HARDWARE A good facial image is required obtained from a camera optimally positioned for the driver facial capture Confidential Power Automation Printed 5 29 2005 ee ee ns k a CUBE Page 19 of 104 Dawa Sewbalak The S 912 USB Video Grabber with Stereo Audio is the ultimate USB video connector featuring built in video and audio ports Its RCA and S Video ports allow you to easily import video from CCTV cameras TV s VCR s camcorders or any other analogue devices The audio port supports stereo audio recording PC Interface USB Rev 1 1 upports camcorders TV VCR video game player analogue cameras etc Video Standard NTSC PAL SECAM Composite analogue or S Video Specifications O Stereo Audio Software Interface Video for Windows TWAIN interface WDM driver File Formats Supported BMP AVI PCX TIF JPG TAG Data Format YUV 4 2 2 YUV 4 2 0 RGB O S Support Microsoft Windows 98 98 SE 2000 ME XP Mac OS 8 6 9 0 Provided by USB port Operating Temperature 0 C to 50 C degrees Non Condensing 10 C to 60 C degrees Non Condensing te l ure Rate 25 fps at CIF 320x240 10 15 fps at VGA 640x480 QCIF 160x120 to VGA 640x480 Hardware Compression O ID 5 jc Oo ia o c lio Q D 5 j Oo z S O o pj c o Minimum System Requirements Intel Pentium II 350 MHz CPU or above 1 Free USB Port 512 MB of RAM Microsoft Windows
137. ust spend to have his her biometric reference template successfully created Equal Error Rate The error rate occurring when the decision threshold of a system is set so that the proportion of false rejections will be approximately equal to the proportion of false acceptances Extraction The process of converting a captured biometric sample into biometric data so that it can be compared to a reference template Confidential Power Automation Printed 5 29 2005 Page 98 of 104 Dawa Sewbalak Failure to Acquire Failure of a biometric system to capture and extract biometric data comparison data Failure to Acquire Rate The frequency of a failure to acquire Failure to Enrol Failure of the biometric system to form a proper enrolment template for an end user The failure may be due to failure to capture the biometric sample or failure to extract template data of sufficient quality Failure to Enrol Rate The proportion of the population of end users failing to complete enrolment False Acceptance When a biometric system incorrectly identifies an individual or incorrectly verifies an impostor against a claimed identity False Acceptance Rate FAR The probability that a biometric system will incorrectly identify an individual or will fail to reject an impostor The rate given normally assumes passive impostor attempts The False Accept Rate may be estimated as FAR NFA NIIA or FAR NFA NIVA where FAR is the false acceptance rate
138. w Conibrol Track 22 1 2 1 Track Mode verify Click on the Track button or select Track from the Control menu to enable add the facial tracking operation Cebisns Cee Matches Ea While operating in Track mode the Discovery application will locate any human face within the video frame place a Tracking Box around the face and follow the face within the frame The Discovery application is able to track up to 16 faces simultaneously In Track mode the application does not attempt to perform identification of the face Classify and Verify operations must be enabled separately by clicking on buttons located below Track The system will remain in Track mode until the tracking operation is disabled Click on the Track button again to disable facial tracking Note that the facial tracking operation cannot be disabled while the Classify or Verify modes are enabled 22 1 2 2 Classify Mode Click on the Classify button or select Classify from the Control menu to enable Classify mode one to many identification Confidential Power Automation Printed 5 29 2005 Page 65 of 104 Dawa Sewbalak While in Classify mode the application will attempt to identify tracked faces from the list of users currently enrolled within the system The results of the classification attempt are displayed in the upper right of the screen in the Confidence Window When a individual is tracked and the classification attempt begins a thumb
139. w user to the system and start the enrollment process When the Add button is clicked the Enter Name window will POGLGTIE appear Fact Hara Ferara Figure 5 The Enter Name window of the Add User function LantMare mca iD Murba pip Mirabe OF ragt 100 onea Enter the User s first name in the First Name box Enter the User s last name in the Last Name box Enter the User s card number in the ID Number box Enter the number of images you wish to collect of the User in the Number of Images box 100 is the default value e Click OK Confidential Power Automation Printed 5 29 2005 JT WTE AFETE T fT gt a ECI Page 67 of 104 Dawa Sewbalak The enrollment process will begin whereby facial images of the user are collected and subsequently added to the FRS database The enrollment progress is shown within the Confidence Window A green progress bar will move further to the right with each image that is collected When the number of images specified in the Number of Images box have been collected the enrollment operation will terminate If the system times out prior to capture of the specified number images or the user leaves the cameras frame for a period of time default 20 seconds the enrollment operation will fail The default value for the timeout condition can be modified in the Enrollment options Section 3 2 3 When the enrollment operation has finished the user will be asked to
140. xisting View cccccececseeeseeeseeeeceeceueeseeeseeecsueesueeseueeseeeseueesaes 87 22 5 3 Opening a New VICW srieccccssdcndsiestowsncdestonccdadensesicnnedsdscuaceatenbucdestininalor daweccssdeness 87 22 54 Saving the Activity LOQ sess ca sectehee cide sseredeinesecdiietauseansitnesedtieosgndied sedauceddciaeed 87 22 95 Loading the ACIVINY LOO sceso EEAO 88 22 6 Modifying Control Button Configuration ccccccccseeeeeceeeeeeeeeeeeeeeeeeseeeeesaeeesenees 88 2 ONT Up oneee e E eel een enee ner eee et 88 22 6 Video SENINGS esee nE R E E E E aaar aE Rai 89 22 9 FRS Server Manager ccccccseccccsesceeceeeeeseeceeeeeeeeeseeceeseeeeessaeeesseusessaueeessaeeeeeags 89 22 9 1 Shutting Down the FRS Central Server cccceceecseeeeeeeeeeeeeeeeeseeeeesaeeeeesaes 90 22 9 2 Restarting the FRS Central Server ccccecccccseeeeceeeeeeeeeeeeeeeeeeeseeeeeseeeeenees 90 22 9 3 Viewing Connected Client Computers cceeccecseeeeeeeeeeeeeeeeeeseeeeeeaeeeeesees 90 22 9 4 Managing FRS Training Servers ccccccccsseecceeeeeeeeeeeeeseeeesseeseesaeeeeeseeeeeeaes 90 22 9 5 Selecting the FRS Database cccccccccceecceceeeeeeseeseeseeeeeseeeeeesaeeeesaaeeeesags 91 22 10 FRS Training Server Managet ccccccccccssscecseseeeseeeceeeeeseeseeceeeeeseeeeesaeeeeeeees 91 23 Biometric Intelligence Overview c cccccceececeeeeseseeceeeeseeeeaees 93 PS ANOT TOOG craen E E E E 93 23 2 Performance Featu
141. y Edenonne functionality by reacting to events and interfacing to i Dede Seng ered Seal Earann peripheral devices Extensions provided with the MCD Fisk Corana Pi 0 Discovery System provide peripheral control for Wiegand tli pall interfaces dry contact relay outputs and mobile wireless computing devices The Extensions tab shown below modifies properties associated with these program extensions Figure 16 Options window Extensions tab Operational parameters concerning an extension are adjusted by highlighting the extension name on the above form and clicking the Properties button The operational features characteristics provided by each extension are described below 22 2 7 1 NCD Relay Control R410 R810 This extension provides logic to toggle dry contact relay outputs on the following events verify success verify failure classification success facial tracking ID card or keypad usage through Wiegand interface Selecting the Properties option for the relay extension enables the following form Figure 17 Relay Options dialog Relay contacts are selected by enabling the checkbox located on the left side of the screen The operation that controls the relay i e verify classify track or Wiegand signal is selected using the option buttons located on the right side of the screen The duration that the relay Confidential Power Automation Printed 5 29 2005 Page 77 of 104 D
142. y on technology is nowhere more real than in the area of biometric surveillance Such surveillance is most effective if the people you are trying to locate are not aware of its use Audit trails left by an individual as he or she uses Casinos car rentals and any other services that require biometric authentication i e possibly any activity that requires the use of a credit card driver s license passport or any other major form of identification could become a significant contribution to intelligence systems T rapiat iai fangs maliy E aiid Confidential Power Automation Printed 5 29 2005 a a eel ee a wane ati GUBE Page 44 of 104 Dawa Sewbalak 16 Privacy discussion Discussion concerning the implementation of large scale biometric systems always includes speculation concerning public attitudes One of the difficulties with what is said about public attitudes on any subject is that interest groups tend to impute their own fears values and biases to the public Most of the interest groups who speak out on the subject of privacy tend to have attitudes that are not friendly to the use of biometrics The danger is that the more those views are repeated the more they will tend to shape public opinion Although there is much talk in the biometric community about the public attitude most who raise the point do so on a very superficial basis There has been little organised dialogue or ongoing discussion concerning the sub
143. ystem from a previously saved export file Select Import Users from the Tools menu and choose the file to be imported The Discovery application will create a new user within the Server database for each user stored within the export file Confidential Power Automation Printed 5 29 2005 Pi BTF REFET A100 THE H Ce CUBE i Page 83 of 104 Dawa Sewbalak 22 3 3 1 Import Options During import of users it is possible that the export file contains one or more user ID numbers that already exist within the current Discovery database To deal with possible duplication of user C Ajeet dupieate CardiD numbers wih urrnatcharl narar IDs during import the Import Options dialog provides two methods To access the Import DO eA Options dialog select Import Options from the File menu kmipori options E Fimer si digh se Cud ramba Figure 24 The Import Options dialog 22 3 3 1 1 1 Reject all duplicate CardID numbers If this option is selected only users with unique ID numbers will be imported If a user ID within the export file conflicts with a user currently enrolled within the database the user will not be imported and the import rejection logged 22 3 3 1 1 2 Reject duplicate CardID numbers with unmatched names If this option is selected imported users with conflicting ID numbers will be discarded if their name cannot be matched to the user currently enrolled within the database If an existing user has both the same ID n

Download Pdf Manuals

Related Search

Related Contents

LIBRETTO ISTRUZIONI INSTRUCTION BOOKLET  真ネコの眼 取扱説明書 - FX情報商材レビュー評価 実践検証ブログ 禁断  soft-air-wds Neusystem  取扱説明書 - 632.53 KB  

Copyright © All rights reserved. Failed to retrieve file