实时人脸识别的一种快速算法  

Fast algorithm of real time face recognition

作　　者：金磊[1] 陈杨[1] 

机构地区：[1]天津大学电子信息工程学院,天津30072

出　　处：《电子测量技术》2007年第5期32-34,共3页Electronic Measurement Technology

摘　　要：当前人脸识别的方向主要是针对静态人脸的单张图像识别,对个体采取多张照片采集,构成一个图片库,然后对捕捉的图像与图片库进行对比来进行人脸识别,鲜有对实时人脸的动态识别。本文提出了一种简单快速的实时人脸识别方法,采用对实时采集的灰度图像进行差分和积分投影处理,再进行极值判断和识别人脸。采用Visual C++对该算法进行编程实现,实验结果验证了算法的有效性。Most current methods of face recognition are based on single static image, sampling multi-pictures to construct a picture data-base,and comparing the pictures which are captured from the picture data-base. However, few of them realize the dynamic recognition in a real time. This paper presented a simple and fast method of face recognition running in a real time. This method used the information of the minus grayscale picture to process by difference and integral projection,judged the extremum and recognizes the human face. At last, this method is implemented in Visual C＋＋ environment, and the experiment results validate the method.

关 键 词：人脸识别 灰度 积分投影 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]