Eye Tracking System Based On FPGA

Eye tracking,as a new type of human-computer interaction technology,has urgent application requirements in many fields such as assisted driving,psychological research,virtual reality,and military.However,existing eye tracking systems rely too much on computer platforms,resulting in large size,high power consumption,and limited application scenarios,which limits their widespread application.This thesis has done research on the implementation of eye tracking system based on embedded hardware.a field programmable gate array(FPGA)is used as the hardware platform to design the eye tracking system based on a head-mounted near-eye camera and image processing technology.With the related upper driver,this system can be used for human-computer interaction with the computer.In the part of system hardware design,the choice of hardware architecture is discussed according to the functional requirements,the device selection of each module is analyzed,and a detailed circuit design scheme is given.With regard to the design of the pupil center positioning algorithm,according to the simple characteristics of the background image collected by the near-eye camera.First the adaptive threshold image binarization is used to extract the rough foreground image including the iris and the pupil,and then the morphological operation and the maximum connected pixel area statistics are combined to remove the Noise were brought by eyelids,eyebrows,eyelashes and corneal reflections,finally resulting in an accurate foreground image containing only the iris and pupil and finding the smallest circumscribed rectangle,according to the concentric circles of the pupil and iris,so use the coordinates of the center of the smallest circumscribed rectangle to approximate the coordinates of the pupil center to realize the pupil center positioning.The system is based on the implementation of the FPGA platform.Verilog hardware description language is used to program the functional modules in the FPGA.First drive the camera module to initialize,get the human eye grayscale video stream through the video data acquisition module,and then drive and control the synchronous dynamic random access memory(SDRAM)to buffer the video stream in frame units,and transfer it to the image.The processing module and the pupil center positioning module process to obtain the pupil center coordinates,and finally send the pupil center coordinate values through the data transmission module in a serial communication(UART)protocol.The results show that the eye tracking system based on FPGA can collect video images of the human eye through the camera and achieve human eye tracking with an accuracy of 95.4% and a response time of 60 fps.It can cooperate with driver software to realize eye movement interaction with the upper computer.