Design of Real-time Eyes Detection System Based on FPGA and DSP

Currently,eyes detection has become the focus in the field of machine vision and pattern recognition. It has widely used in the applications of face detection and recognition,driver fatigue detection, human-computer interaction and video surveillance.These applitions require that eyes detection has the high real-time and accuracy performance, so looking for the system with high speed and accuracy is of great significance.In this paper,the eyes detection system was designed on the hardware platform of FPGA and DSP based on the principle that the human eye pupil has different reflectivity with different wavelengths of near-infrared.The system realized the function of eyes detection through the process of difference,segmentation,integral projection,connection domain and improved complexity alogorithm on the image collected by image sensor.In the process of system development,first,hardware circuit was designed,including lamp board collecting circuit and core circuit of FPGA and DSP processor.Then,software module was designed,including image acquisition module, the frame memory module, image processing module and display module.Finally, the system of hardware module and software module was co-debugged.The system combined video image acquisition,frame storage, processing and display through allocating the tasks resonably according to the characteristics of FPGA and DSP. The speed of image acquisition is up to 60 frames per second and the speed of image processing is up to 30 frames per second. This system can meet the demand of real-time. The system has the characteristics of small size, low power consumption,great anti-interference ability, expanding functions easily and good broad prospects.