A real-time human eye pupil tracking system using FPGA

Video-based eye tracking is a valuable method in a variety of real-time applications such as diagnostic equipment, ophthalmology, assistance for wheelchair people, and so on. The pupil tracking system is a real-time implementation with a difficult task when pupil movement, occlusion of eyelids, and eyelashes present some challenges. To effectively address all aspects of these real-time problems, we present a robust pupil tracking system for vision based on black box technique in FPGA. Previous Circular Hough transform methods are contrasted with a proposed pupil detection method that improves resource utilization in FPGA synthesis. The proposed system includes a digital CMOS (Complementary Metal Oxide Semiconductor) camera for low power considerations, LPDDR (Low Power Double Data Rate) external memory, the histogram equalization technique for contrast enhancement, the hysteresis threshold for segmentation, and the black box technique for finding pupil center coordinates to track the eye, all of which have been implemented on the Xilinx Spartan6 (xc6slx45) FPGA Board. The experimental results demonstrated the accuracy and effectiveness of implementing a proposed real-time-based pupil tracking system over existing tracking systems. The proposed technique used the 30fps with 640x480 resolutions of the input video frames and FPGA resource utilization of the individual modules is summarized.