人脸自动追踪与识别系统的设计与实现  

Design and Implementation of an Automatic Track and Recognition System of Human Faces

作　　者：肖南峰[1] 郑宇[1] 

机构地区：[1]华南理工大学计算机学院,广东广州510640

出　　处：《华南理工大学学报（自然科学版）》2004年第9期1-5,共5页Journal of South China University of Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目 (6 0 375 0 31) ;广东省自然科学基金 (重点 )资助项目 (36 5 5 2 )

摘　　要：为快速和准确地识别有关人员的身份 ,开发了一个具有 5个自由度的人脸自动追踪与识别系统 .该系统采用人脸肤色模型结合相似度及人脸轮廓来检测和定位人脸 ,并利用嵌入式隐Markov模型对人脸的特征进行提取和识别 .实验结果表明 ,该系统不仅能自动追踪、定位和识别人脸 ,而且能实时地将识别结果与人脸数据库进行比对 ,从而迅速地获得被识别人员的有关身份信息 .该系统可用于建筑物入口的监控和刑事侦察等场合 .An automatic track and recognition system of human faces with five freedom degrees was developed to rapidly and correctly recognize a person. In this system, the skin-color model combined with the comparability degree and the face figure was adopted to detect and segment a human face, and the embedded hidden Markov model was applied to the extraction and recognition of face characteristics. The experimental results indicate that the system can not only track, locate and recognize a human face automatically, but also confirm an identity on time by comparing the recognized images with those in the face database. The proposed system can be used to reconnoitre criminal cases and monitor the entrances of buildings.

关 键 词：人脸检测 追踪 识别 隐MARKOV模型 图像处理 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]