基于FPGA平台的实时人眼检测系统设计

摘要

The near-infrated source and image difference methods were used to design the eyes detection system based on FP- GA. Through the image difference processing, a difference image with bright pupils which takes the face as the main body can he obtained. Then, through gray stretching, image segmentation and intergal projection on difference image, the location of the face can be obtained, and then the eyes detection can be implemented through improved complexity algorithm in the de- tected face region. After experience, the processing speed of the system can achieve 30 frames persecond, which meets the requirement of real-time performance. Meanwhile, the method of near-infrared source can eliminate the effection from other sources, so the system＇s robustness can be graranteed.%采用近红外光源与图像差分相结合的方法在现场可编程逻辑阵列（FPGA）平台上实现了人眼检测系统。利用图像差分可以获得具有亮瞳孔效应、以人脸为主体的目标图像，通过对其进行灰度拉伸、分割、积分投影可以得到人脸位置，在检测到的人脸区域利用改进的复杂度算法实现了人眼检测。经测试，系统的处理速度可达30帧／秒，满足实时性检测的要求。基于近红外光源的方法消除了外界光源的干扰，从而保证了系统的鲁棒性。