Facial Recognition General-Purpose Access Control Device Alex
Contents
1. Facial Landmark Detector Communications Process values tx doGet doPost Web server Process Reading Image Ima ge Function storage SRAM Get Image Image retrieval Board Control HW retrieved Process Fig 1 Process interaction Test Plan Software Image analysis Test of facial recognition in example images on laptop Success Test of facial recognition software on Altera DE2 board Failure not enough RAM to fit requirements Test of facial recognition of images pulled from web server Success Compression Compression abandoned due to time Web Server 1 Serve sample web page Success 2 Serve sample image in GET response Success avg 55kB s speed 3 Respond to POST requests with sample reply Success 4 Convert sample raw RGB image data to Bitmap image Success 5 Trigger image read from camera and display as bitmap Success avg 55 kB s speed 6 Serve auto refreshing web page Success 7 Serve Identified Person web page Success 10 8 Serve unidentified person web page Success 9 Search database with sample facial measurements and identify person Success 10 Store name and sample facial measurements from unidentified person STORE POST reguest Success 11 Test STATUS REQUEST POST response with facial recognition server Success 12 Respond to FACE FOUND reguest from facial recognition server Success 13 System will auto refresh until facial recognition server finds a face i2. 20 vhd niosll system burst 21 vhd niosll system burst 22 vhd niosll system burst 3 vhd niosll system burst 4 vhd niosll system burst 5 vhd niosll system burst 6 vhd niosll system burst 7 vhd niosll system burst 7 vhd bak niosll system burst 8 vhd niosll system burst 9 vhd niosll system clock 0 vhd niosll system clock 1 vhd niosll system clock 10 vhd niosll system clock 11 vhd niosll system clock 12 vhd niosll system clock 13 vhd niosll system clock 14 vhd niosll system clock 15 vhd niosll system clock 2 vhd niosll system clock 3 vhd niosll system clock 4 vhd niosll system clock 5 vhd niosll system clock 6 vhd niosll system clock 7 vhd niosll system clock 8 vhd niosll system clock 9 vhd niosll system generation script niosll system inst vhd niosll system OO Dei niosll system sim nvram controller vhd nvram controller vhd bak nvram controller 0 vhd nvram controller hw tcl nvram controller hw tcl onchip_memory2_0 hex onchip_memory2_0 vhd onchip_memory2_inst hex onchip_memory2_inst vhd sdram vhd seven_seg_pio vhd sopc add gip file tcl sopc builder log txt sram vhd sram controller vhd sram controller hw tcl sram controller hw tcl stp1 stp switch vhd 17 sysid vhd sys_clk_timer vhd transfer_FIFO cmp transfer_FIFO qip transfer_FIFO vhd VGA_Adapter vhd VGA_Adapter vhd bak VGA_Adapter_hw tcl VGA_Adapter_inst vhd VGA PLL gip VGA PLL v videoln 48 vhd videoln vhd videoln vhd bak videoln copy vhd Videoln Co
3. A U minei system AO H alipl inst c0 out altpll_inst_cl_out altpll_inst_c2_out sram_s0_data 15 0 locked from the altpl inst T p Teka om he GE rom he dota BOTE RRE FG D rr ENET_INT_to_the_dm9000a_inst ENET_CMD_from_the_dm9000a_inst err S BEGUE irom the dma ENET CS N from the dm9000a inst X gt gt ENET CS N from the OSU inst i ENET DATA to_and_from_the_dm9000a_inst 15 0 Z BET DAYA iO and fromthe RE sii ENET RD N from the dm9000a inst ENET RST N from the dm9000a inst ble C gt ENET_RST_N_from_the_dm9000a_inst EEL ME N from the dm9000a inst ENEE WA N an OS SI LCD E from_the_led display y LCD RS from the bd display a ES LCD_RS_from_the_kd_dsplay i LCD RW trom the_ed_dsplay a LCD data to and from the bi display 7 0 out port from the led pio 7 0 Se addr from the sdram 11 0 EEA S zs ba from_the_sdram 1 0 kada EE fromthe sami fT zs cas n from the sdram E zs cke from the sdram E zs cs n from the sdram E 25 d to and from the sdram 15 0 eru o and from he sdram is 0j EE zs_dqm_from_the_sdram 1 0 ped T 25 dom rom ihe AST 2s ras n from the sdram aS E e Tas fromthe sdam p zeun n from the sdram bars ZE EEK ZTU ld out port from the seven seg pio 15 0 coe address export from the sram 17 0 coe chip enable export n from the sram coe lower byte export n from the sram coe_output enable export n from the sram E Ham 08 output enable expo
4. SRAM and serve the found image as a Bitmap file If it receives a GET request for found face bmp it will serve up the image data that is already on the SRAM as a Bitmap image If it receives a GET request for facefinder html it will serve up a dynamically generated html page If no face has been found this page will be an automatically refreshing simple page with no information If a face has been found and the person has been identified the page will display the image the name of the identified person and an html form with a list of actions to take in response unlock door release hounds etc If a face has been found but not identified the page will include a form where a name can be entered by the user which will be saved with the measurements in the list of known faces The web server can also respond to a number of POST requests If it receives a STATUS REQUEST request it will return NOT_READY if the dynamic page is either not being viewed or is currently showing a face and waiting for the user to refresh the page If the page is being viewed but is in automatic refresh mode i e no face has been found it will return READY FOR FACE If it receives a FACE_FOUND request where hold the facial measurements it will extract those measurements and signal that a face has been found It will then attempt to identify the person in the image by the measurements and if the measurements are in the list of known faces the system will
5. signal that the person has been identified If the system receives a STORE Post request with containing the facial measurements of the face currently being displayed it will read the name contained in the post reguest and save it with the facial measurements in the list of known faces There are also Post reguests for the actions the user can take which trigger board functions like sweeping the LEDs SWEEP and writing text to the LCD display LCD Board Control process This process will monitor for messages sent from the web server to perform basic board functions such as turning LEDS on or off displaying text on the LCD or sending pixel data to the Ethernet controller Communications server This is a java program running on a connected computer We had intended for this to be contained on the board but it reguired more resources than the board could provide This program will periodically poll the Web server process with STATUS REGUEST POST requests to determine whether or not the system is ready to analyze image data Once it is ready the process will send a GET reguest to the webserver for find face bmp storing the image data and saving it as image bmp The process will then wait for values txt to be created by the landmark detector once it has been detected the float values will be read in and stored and values txt will be deleted If all 6 reguired values are there they will be used to construct a doPost reguest send
6. Facial Recognition General Purpose Access Control Device Alex Newcomb newcomb Qualberta ca Tom Stefanyk tom stefanyk ualberta ca Preferred Lab Day Wednesday Other lab days Tuesdays and Thursdays after 3pm Summary Detecting and recognizing a face from input video stream and displaying on a web server Table of Contents Abstract Functional Reguirements Design and Description Parts List I O Signals Datasheet Software Design Test Plan References Appendices Quick Start Guide Future Work Hardware Documentation Software Documentation ea ea ON Abstract The facial recognition device is designed with automated access control in mind The device uses a camera to take in a video stream of which still images will be periodically stored Once algorithms detect the presence of a face within such a still image key measurements will be taken of the facial features and compared to a database of known faces The image will be converted to a simple image format and made available via an on board web server hosting a simple website which will display the image along with the name of the person if recognized Additionally the website will contain a list of actions the person can take Functional Requirements Original Requirements The device must take video input from a video camera or similar device interfaced via the composite video input The image is to be saved both in a compressed JPEG color format and as a simple g
7. In FIFO v Videoln Core Video In FIFO v bak Videoln Core ideo In FIFO inst v Videoln Core Video In FIFO old v Program Code C Java Assembly etc In directory Facial Recognition Server face detector0 detector c face detector0 facedetector c face detector0O structs h face detector0 haarcascade eye xml face detector0 haarcascade mcs mouth xml face detector0 haarcascade mcs nose xml Face Recog Server src server java In directory webServer dm9000a HAL inc dm9000a UCOSTIl src software webServer alt error handler c software webServer alt error handler h software webServer create this app software webServer http c This handles the HTTP GET and POST responses it keeps track of facial data serves up the dynamic web page retrieves image data from SRAM and converts to Bitmap and holds and searches the list of known faces software webServer http h software webServer image h software webServer Makefile software webServer network_utilities c 19 software webServer srec flash c software webServer web_server c software webServer web_server h software webserver_bsp alt_sys_init c software webserver bsp libucosii bsp a software webserver bsp linker h software webserver bsp linker x software webserver bsp Makefile software webserver bsp system h software webserver bsp drivers inc software webserver bsp drivers src software webserver bsp HAL inc software webserver bsp HAL src software webserver b
8. e DE2 board and connect the laptop to the DE2 board via Ethernet cable 10 Build and run the face_detector0 project on Eclipse on the laptop 11 Build and run the Face_Recog_Server project on Eclipse on the laptop 12 Open your browser of choice and enter http 196 254 70 234 facefinder html to start the face recognition polling GO Sl GO GO b GA b ek EO Appendix B Future Work e Implement rudimentary emotion recognition eg smiling happy frowning sad furrowed brow angry e Humorous automated image modifications eg change color of irises add silly mustache et cetera Implement more advanced facial recognition algorithm Provide audio notification from the board once a face has been recognized Add in JPEG compression of the image file to allow for speedier operation Include audio video streaming capability Appendix C Hardware diagram Nios II processor Composite Video In ADV7181 Video Camera Bitmap SRAM generation Ethernet Network DM9000A Client computers 13 Schematics y 1 1 EU E S A D 1 1 1 1 1 1 i 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1 1 1 1 1 i 1 i 1 i 1 i 1 i 1 i 1 1 1 i 1 i 1 1 1 i 1 i 1 1 i i 1 i 1 1 1 i 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1 i 1 1 i 1 i
9. from the dm9000a inst command line 1 wire ENET CS N from the dm9000a inst chip select active low 1 wire ENET DATA from the dm9000a inst data lines 16 wires ENET INT from the dm9000a inst interrupt line 1 wire ENET RD N from the dm9000a inst read enable active low 1 wire ENET RST N from the dm9000a inst reset active low 1 wire m ENET WR N from the dm9000a inst write enable active low 1 wire o Clock source m clk_0 50 Mhz clock 1 wire m clk_27MHz 27 Mhz clock 1 wire 3 Board to Off board Peripherals o Ethernet cable between Ethernet controller and other computer o Composite video input from camera 4 Off board Peripherals to World o Power adapter to Altera DEZ board 9V 1 3A o Power adapter for camera 11V 1A Datasheet y 1 i 1 EU E S A D 1 1 1 1 1 1 i 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1 1 1 1 1 i 1 i 1 i 1 i 1 i 1 i 1 1 1 i 1 i 1 1 1 i 1 i 1 1 i i 1 i 1 1 1 i 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1 i 1 1 i 1 i A U niosll_ system AO H alipl inst c0 out altpll_inst_cl_out altpll_inst_c2_out sram_s0_data 15 0 locked from the altpl inst T p Teka om he GE rom he dota BOTE RRE FG D rr ENET_INT_to_the_dm9000a_inst ENET_CMD_from_the_dm9000a_inst err S BEGUE irom the dma ENET CS N f
10. ified and an HTML form containing a list of actions the user can take Examples of actions are Unlock the door Send a message or Release the hounds In our case these choices result in different board actions such as a sweep of the on board LEDs display of the message on an LED display or changing of the on board 7 segment display If the person in the image is not identified the web page also prompts the user to enter the name of the person who s name and facial measurements are then stored in the list of known faces on the device To ensure that the image on the screen is the one in which the face is recognized the facial recognitions stops requesting images once a face has been identified and polls the device to see if it is ready for a new image If the web page is refreshed the device will respond that it is ready for a new image Parts List Terasic DE2 development board Website http www altera com education univ materials boards de2 unv de2 board html User Manual ftp ftp altera com up pub Webdocs DE2_UserManual pdf JVC Digital Video Camera GR DVL510U 0 already owned Website http support jvc com consumer product jsp archive true amp pathId 26 amp modelld MODLO026470 User Manual http resources jvc com Resources 00 00 98 LYT0722 001A pdf Outputs Composite video Acer Aspire 5738 Laptop 0 already owned Specification http support acer com acerpanam notebook 2009 acer aspire Aspire5738 A
11. ing those values back to the Web server process signifying a face has been successfully recognized If not the program will request another image via the doGet request and repeat the cycle until a face has been successfully recognized Facial Landmark Detector This process will wait for the Communications process to create image bmp at which point it will read it into a data structure and delete the file A greyscale copy will be made of the color image and both images will be resized from 320x240 to 640x480 Next the resized greyscale image will be equalized to increase contrast and the XML file containing the eye patterns to search for will be loaded into memory A function will then sweep through the image to locate matches for the eye pattern The coordinates of the matching points will be saved and circles will be placed over the points on the resized color image This process will be repeated with XML files containing nose and mouth patterns For each of those sweeps if there are too many points found only the point most likely to correspond to the desired facial feature will be kept Once the three sweeps have been made the distance between each point will be calculated divided by the distance between the first two points which will be the left and right eye upon a successful sweep and stored as single precision float values A file called values txt will be created and all the values written to it image bmp
12. n rom coe upper byte export n from the sram hai EE S Sea upper byte export A from ihe coe write enable export n from the sram BDB eri essa Out port from the seven seg pio15 0 i Se EO OD enable export n irom the sram R in_port_to_the_switch address_to_the_ext_flash 21 0 A read n to the ext flash gt read_n_to_the_ext_flash i select_n_to_the_ext_flash gt H tri_state_bridge_flash_data 7 0 per write n to the ext flash bire coe TD DATA export to the videoln inst 7 0 coe I2C SCLK export from the videoln inst E coe_TD_HS_export_to_the_videoln_inst coe I2C EDA export to and from the videoln inst E _I2C_SDAT 6 A coe_TD_VS_export to_the_videoln_inst coe_TD_RESET_export_ from the videoln inst Bi AS cos YO RESET expan trom ezdu bei beai ei 4 out port led from me videoln inst out por led from the videoin T BE a tae edge Ma 1 D i H 1 inst H LJ 14 Appendix D Software Diagram image bmp Facial Landmark Detector Communications Process doPost Web server Process Reading Image Do Board Function Bitmap Image storage SRAM Image retrieval Board Control HW retrieved Process Source code Verilog VHDL In directory webServer altpll inst vhd altpll inst vho bigfatFIFO cmp bigfatFIF O gip bigfatFIFO vhd clock source 27 vhd cpu ocp cp
13. n image served by web server find returned measurements in database show web page repeat on page refresh Success 14 Writing to SRAM Failure due to SRAM clock speed being higher Board Control Test that each function results in the desired response from the board Hardware Ethernet Test that connection detection is working i e lights turn on with cable is plugged in Success Test that TCP IP communication is working ping responses are correct Success Image Retrieval Test image data conversion from YCbCr to RGB by outputting directly to VGA Success occasional discoloured lines appearing in image Test writing image to SDRAM Failure Test writing image to SRAM Success some discoloured lines but less noticeable Results of Experiments Originally we had intended to use 640x480 JPEG images but due to lack of space on SRAM and a not having enough time to implement JPEG compression we ended up using 320x240 RGB image data which is converted to Bitmap format on the fly Despite our best efforts we could not fit the facial recognition program onto the DE2 board so we wrote a program to do the facial recognition on the computer connected to the board leaving the board to handle image generation and web serving This required setting up a 2 way communication system between the web server and the facial recognition program 11 References 1 Sample Webserver amp DM9000A driver from Terasic samples at http www te
14. rasic com tw cgi bin page archive pl Language English CategoryNo 538No 308PartNo 4 2 Video input example from University of Toronto ECE 241 website at http www eecg toronto edu jayar ece241_08F AudioVideoCores vin vin html 3 Facial recognition via Eigenfaces method from Open source examples at http www face rec org source codes 4 SimpleSocketServer by user hasil86 referenced to help setting up web server on DE2 board http alteraforums com forum showthread php t 240208page 7 5 OpenCV library and wiki at http opencv willowgarage com wiki 6 JPEG compression code found at http www ijg org 7 Javascript auto refresh http www quackitcom javascript javascript_refresh_page cfm 12 Appendix A Quick Start Guide If the laptop does not have the OpenCV library compile and install it at C OpenCV2 3 Compile the webserver lt extension gt Quartus project Connect DE2 board and refresh the JTAG daemon Program webserver sof or webserver pof to the DE2 board Open Nios II Software Build Tools for Eclipse via the SOPC builder Generate the BSP for the webserver_bsp project Build the webserver project Run the webserver project as Nios II Hardware On the laptop set the Local Area Connection IP address to 196 254 70 xxx where xxx does not equal 234 the subnet mask to 255 255 255 0 and the default gateway to 196 254 70 1 9 Connect the composite video output of the camera to th
15. re Videoln Core Mod Videoln Demo Mod vhd Videoln Demo Mod vhd bak videoln hw tcl videoln hw tcl bak videoln hw tcl videoln inst vhd videoln no FIFO vhd Video input fifo cmp Video input fifo gip Video input fifo vhd Video In FIFO gip Video In FIFO mod cmp Video In FIFO mod gip Video In FIFO mod vhd webServer asm rpt webServer cdf webServer done webServer dpf webServer eda rpt webServer fit rpt webServer fit smsg webServer fit summary webServer flow rpt webServer jdi webServer map rpt webServer map summary webServer merge rpt webServer pin webServer pof webServer pow rpt webServer pow summary webServer qpf webServer qsf webServer qsf bak 18 webServer sof webServer sta rpt webServer sta summary webServer vhd webServer vhd bak avconf avconf v avconf I2C Controller v greybox_tmp greybox_tmp mg3ms v greybox_tmp greybox_tmp mglks v greybox_tmp greybox_tmp mgons v Videoln Core Altera UP ITU R 656 Decoder v Videoln Core Altera UP Video In Buffer v Videoln Core Altera UP Video In Deinterlacer v Videoln Core Altera UP Video In Deinterlacer v bak Videoln Core Altera UP Video In Resize v Videoln Core Altera UP Video In Resize v bak Videoln Core Altera UP YCrCb 422 to 444 Converter v Videoln Core Altera UP YCrCb to RGB Converter v Videoln Core Altera UP YCrCb to RGB Converter v bak Videoln Core Dual Clock FIFO v Videoln Core Video In v Videoln_Core Video_In v bak Videoln_Core Video_ln_FIFO qip Videoln Core Video
16. reyscale image Said device must periodically check the saved image the greyscale version for the presence of a face The device must also provide an web based interface via wired Ethernet which will only display images once the device has detected a face Said interface must display identifying the face as either a specific known person or as an unknown person the interface must also provide several actions for the user to choose in response Performance The finished device met most requirements but some trade offs were needed Due to the size and requirements of the facial recognition algorithm this functionality was moved from the device to a computer on the network which we call our facial recognition server The device serves images from the camera to the facial recognition server which sends the facial measurements of any recognized faces Also the JPEG compression functionality was cut due to time constraints which slows the web site loading considerably Design and Description The device is on a network with a facial recognition server and a client computer It connects to this network via the on board Ethernet device In our example the facial recognition and client computers are the same The device serves a dynamic i e created on the fly web page to the client computer and keeps track of whether the web page is being viewed The web page has an auto refresh function implemented with a simple javascript on load func
17. rom the dm9000a inst X gt gt ENET CS N from the OSU inst i ENET DATA to_and_from_the_dm9000a_inst 15 0 Z BET DAYA iO and fromthe RE sii ENET RD N from the dm9000a inst ENET RST N from the dm9000a inst ble C gt ENET_RST_N_from_the_dm9000a_inst EEL ME N from the dm9000a inst ENEE WA N an OS SI LCD E from_the_led display y LCD RS from the bd display a ES LCD_RS_from_the_kd_dsplay i LCD RW trom the_ed_dsplay a LCD data to and from the bi display 7 0 out port from the led pio 7 0 Se addr from the sdram 11 0 EEA S zs ba from_the_sdram 1 0 kada EE fromthe sami fT zs cas n from the sdram E zs cke from the sdram E zs cs n from the sdram E 25 d to and from the sdram 15 0 eru o and from he sdram is 0j EE zs_dqm_from_the_sdram 1 0 ped T 25 dom rom ihe AST 2s ras n from the sdram aS E e Tas fromthe sdam p zeun n from the sdram bars ZE EEK ZTU ld out port from the seven seg pio 15 0 coe address export from the sram 17 0 coe chip enable export n from the sram coe lower byte export n from the sram coe_output enable export n from the sram E Ham 08 output enable expo n rom coe upper byte export n from the sram hai EE S Sea upper byte export A from ihe coe write enable export n from the sram BDB eri essa Out port from the seven seg pio15 0 i Se EO OD enable export n irom the sram R in_port_to_the_switch addres
18. s created it will be analyzed for signatures that identify locations of facial features by a facial recognition program running on a computer on the same network as the web server If such signatures are found and a face is recognized the relative facial measurements relative to the distance between the eyes E will be sent to the web server where they will be compared against a list of known faces If the relative distances of the facial features of the captured face are close enough to that of an entry in the list the system will identify the face as belonging to a known person In this case close enough means each measurement is within 0 2E of the known face If the facial features do not match those in the database the system will note that the face belongs to an unknown person Regardless of whether the person is identified if a face is recognized an update will be prepared for the web server consisting of a picture of the identified face and the name of the identified person if applicable If a face is not recognized in the image taken from the composite input the image will be discarded and the system will fetch a new image Processes Web Server This will be based off of the example provided by Altera and included with the Nios II IDE It monitors for GET and POST requests from clients If it receives a GET request for an image find_face bmp it will trigger the system to retrieve an image from the video input stream save iton
19. s_to_the_ext_flash 21 0 A read n to the ext flash gt read_n_to_the_ext_flash i select_n_to_the_ext_flash gt H tri_state_bridge_flash_data 7 0 per write n to the ext flash bire coe TD DATA export to the videoln inst 7 0 coe I2C SCLK export from the videoln inst E coe_TD_HS_export_to_the_videoln_inst coe I2C EDA export to and from the videoln inst E _I2C_SDAT 6 A coe_TD_VS_export to_the_videoln_inst coe_TD_RESET_export_ from the videoln inst Bi AS cos YO RESET expan trom ezdu bei beai ei 4 out port led from me videoln inst out por led from the videoin T BE a tae edge Ma 1 D i H 1 inst H LJ Software Design The interface of the system is a web page on any computer on the system that regularly polls the web server on the system for updates using client side javascript An update will consist of an image of an identified face and the name of an identified person The web page will also have a form that can be submitted to activate various security features in response to an identified face which will likely be represented by LED responses on the DE2 board or to reguest a new image If an update is received the web page will stop polling the web server until the user submits an action On the server side the system will convert the signal from the on board composite video into 16 bit RGB image data This image data is served up by the web server as a bitmap image Once an image i
20. sp iniche inc software webserver bsp iniche src software webserver_bsp UCOSll inc software webserver_bsp UCOSII src 20
21. spire5738sp2 shtml Input Output Ethernet crossover cable I O Signals format signal name signal description signal size 1 Microprocessor to custom HDL o Control signals to video in logic m avm vidln address Avalon Memory Master address 26 wires avm_vidln_write Avalon Memory Master write enable 1 wire avm_vidln writedata Avalon Memory Master data line 16 wires avm_vidln_waitrequest Avalon Memory Master wait request 1 wire avm_vidln_byteenable Avalon Memory Master byte enable 2 wires avm_vidln_arbiterlock Avalon Memory Master arbiter lock 1 wire avs control address Avalon Memory Slave address 4 wires avs control write Avalon Memory Slave write enable 1 wire avs control writedata Avalon Memory Slave data line 8 wires avs control byteenable Avalon Memory Slave byte enable 1 wire 2 FPGA to Board O Data and control lines to from TV encoder m coe TV DATA data line 8 wires m coe_TD HS Horizontal sync 1 wire m coe TD VS Vertical sync 1 wire m coe TD RESET reset 1 wire Control lines to from I2C unit m coe I2C SCLK I2C clock 27Mhz 1 wire m coe_I2C_SDAT data line 1 wire Data and control lines to CFI Flash m address to the ext flash address lines 22 wires m tri state bridge flash data data lines 8 wires m read_n_to the ext flash read enable active low 1 wire m select n to the ext flash chip enable active low 1 wire m write n to the ext flash write enable active low 1 wire Reset switch m in port to
22. the switch reset 1 wire Data and control lines to SRAM m coe address export from the sram address lines 18 wires m sram SU data data lines 16 wires m coe chip enable export n from the sram chip enable active low 1 wire m coe lower byte export n from the sram lower byte enable active low 1 wire m coe upper byte export n from the sram upper byte enable active low 1 wire m coe output enable eport n from the sram read enable active low 1 wire m coe write enable export n from the sram write enable active low 1 wire Seven segment control lines m out port from the seven seg pio control lines 16 wires Data and control lines to from SDRAM m zs addr from the sdram address lines 12 wires zs ba from The sdram byte enable lines 2 wires zs cas n from the sdram Column Address Strobe active low 1 wire zs cke from The sdram clock enable 1 wire zs cs n from the sdram chip select active low 1 wire zs d t and from the sdram data lines 16 wires zs ras n From the sdram row address strobe active low 1 wire zs we n from the sdram write enable active low 1 wire LED control m out port from the led pio led control line 1 wire LCD Display control lines m LCD E from the Icd display enable line 1 wire m LCD RS from the Icd display row select 1 wire m LCD RW from the lcd_dislay read write select 1 wire m LCD data to and from the lcd display data lines 8 wires Data and control lines to from Ethernet controller m ENET CMD
23. tion On request from the facial recognition server via HTTP GET request the device takes video input from a camcorder or other such video device via the composite in input The onboard composite video reads in the video in YCbCr format The image data is converted to 16 bit RGB format The image data is placed in SRAM as a 320x240 resolution image The device translates the image data to Bitmap image format and sends this image to the facial recognition server The server sweeps the image using patterns of pixels that correspond to specific facial features eyes nose and mouth Should the server find all of the aforementioned features the facial recognition server sends the device a signal that a face is present in the current image along with the measurements of the distance between the facial features These measurements are all normalized by the distance between the eyes so a face s measurements should be close to the same regardless of how far away the face is from the camera provided the server can still recognize the face If the server is unable to find all the points it will reguest an additional image from the device and repeat the sweep The device compares the facial measurements against a list of known faces to try and identify the person in the image If a face is contained in the current image the next time the client computer reguests the web page the page will display the image along with the name of the person if ident
24. u sdc cpu vhd cpu_bht_ram mif cpu_dc_tag_ram mif cpu_ic_tag_ram mif cpu_inst ocp cpu_inst sdc cpu_inst vhd cpu inst bht ram mif 15 cpu inst dc tag ram mif cpu inst ic tag ram mif cpu inst jtag debug module syscIk vhd cpu inst jtag debug module tck vhd cpu inst jtag debug module wrapper vhd cpu inst mult cell vhd cpu inst ociram default contents mif cpu inst oci test bench vhd cpu inst f ram a mif cpu inst rf ram b mif cpu inst test bench vhd cpu tag debug module sysclk vhd cpu IO debug module tck vhd cpu IO debug module wrapper vhd cpu_mult_cell vhd cpu_ociram_default_contents mif cpu_oci_test_bench vhd cpu_rf_ram_a mif cpu_rf_ram_b mif cpu_test_bench vhd dm9000a dm9000a vhd dm9000a_hw tcl dm9000a hw tcl dm9000a_inst vhd high_res_timer vhd jtag_uart vhd Icd_display vhd led_pio vhd niosll_system bsf niosll_system html niosll system ptf niosll_system ptf 8 0 niosll_system ptf bak niosll system ptf pre generation ptf niosll system gip niosll system sopc niosll_system sopcinfo niosll_system vhd niosll_system_burst_0 vhd niosll system burst 1 vhd niosll system burst 10 vhd niosll system burst 11 vhd niosll system burst 12 vhd niosll system burst 13 vhd niosll system burst 14 vhd niosll system burst 15 vhd niosll system burst 16 vhd niosll system burst 17 vhd niosll system burst 18 vhd niosll system burst 19 vhd 16 niosll system burst 2 vhd niosll system burst
Download Pdf Manuals
Related Search