人脸检测跟踪与特征点定位系统  

Face detection tracking and feature point positioning system

作　　者：何英杰[1] 李国新[1] 

机构地区：[1]长安大学信息工程学院,陕西西安710064

出　　处：《电子设计工程》2012年第8期189-192,共4页Electronic Design Engineering

摘　　要：实现的人脸检测跟踪与特征点定位系统,基于VC++6.0开发平台,使用opencv作为开发工具,有效缩短了系统的开发时间。首先,本系统采用adaboost算法进行人脸检测,通过合理的特征模板的选择实现了人脸的实时检测;其次,人脸跟踪模块选用camshift算法,利用人脸检测模块生成的人脸坐标传递给跟踪模块,实现人脸的自动实时跟踪,同时建立多个camshift跟踪器对多人脸进行跟踪,并有效地解决了人脸遮挡的问题;最后,通过ASM(active shapemodel)算法实现了实时人脸特征点定位。实验结果表明该系统实现的人脸实时检测跟踪及特征点定位,效果明显,可以作为表情分析和情感计算、视频人脸识别开发的基础。Face detection and tracking and feature point locating system of this article,based on VC＋＋6.0 development platform,using OpenCV as development tools,effectively shortens the development time for systems.Firstly,The system adopts the adaboost algorithm for face detection,implements face real-time detection by means of a reasonable feature template selection.Secondly,the face tracking module selects camshift algorithm,the article uses face detection generating face coordinates passing it to the tracking module,realizes automatic face real-time tracking,and tracks multiple faces by using multiple camshift Tracker,and effectively solves the problem of face block.Finally,the system accomplishes real-time facial feature point location algorithm by ASM（active shape model）.The results state that the effect of the system of real-time detection and tracking of facial feature point location is obvious,the algorithm can be used as the basis of expression analysis,emotion computing,the development of video face recognition.

关 键 词：ADABOOST算法 CAMSHIFT算法 ASM算法 人脸跟踪 特征点定位 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]