Eye tracking system based on SOPC

Eye tracking is one of the potential techniques for the future human-computer interface. In order to promote an eye tracking device that take account of high performance, miniaturization, low cost and power consumption, we implemented the Purkinje eye tracking algorithm into a single FPGA chip with the SOPC (System on Programmable Chip) technology. The image capture, Adaboost algorithm based eye detection and RANSAC based ellipse-fitting were promoted as hardware module in this paper. And a NIOS II on-chip processor took charge of USB protocol and task scheduling. After the verification on a low-cost FPGA chip, the eye tracking system was able to detect the gaze direction and deliver the gaze information through USB interface. The experiment results show that the detection speed was 11 fps under the video stream with a resolution of 640×480, achieving the real time performance. The system integrated the eye tracking algorithm and had the advantages of speed, power consumption, flexibility and cost.