计算机人脸识别技术的研究-信号与信息处理专业论文.docxVIP

计算机人脸识别技宋的研究Study 计算机人脸识别技宋的研究 Study ofthe technology ofhuman face recognition with computer Abstract m technology of human face recognition is active subject in the field of pattern recognition There broad application西the fields ef laws．business and safety systems For the particularity ofhuman face images，face fecog珏i畦on wi氇a computer is very difficult problexn and there a#still many work．s do before such technology be used诫l曲．Wi氇 the devolpment of the society,the application of face recognition systems will be wilder and brings much challenge thereseachers． The buman face recognition system is kind of pattern recognition gy30em based Lrlformation processing．It be divided into two parts：feature extraction and pattern classification．The first part is find of features that CaD．represent the images丹om different persons；曲e second classifies the features got flom the first part．The performance ofthe system depends o珏both ofthe parts． For the first part of the system．the methods W蠢in the thesis principle component analysis(PCA)，Fisher linear discfiminant谬LD)and independent componenl analysis which statistical methods and widly used谊signoa processing and pattem recognition． For the second part ofthe system，SVM multi-classifers are applied to elassi母the features from the first part．13ecause SVM Can only slove two classes problem，for K classes problem．a of SVM must be used HOW combine of SvIⅥinto efficient multi． classifier is probIem worthy of doing much research WOrk．硝thought there kinds of traditionalwaystocombinea∞}《SVM。accodmgtotheanalysisinthethesisnone ofthem拓 efficient for K classes problem．强e method of ellx)r-conecthlg output codes疆COc)is introduced combine SVM into multi—classifier in this thesis，with this metho盎we∞n medinm number of SVM efficient multi-classifier．Because of the predefined ability of correcting,the multi—calssifer give proper result thou曲 SVM give wrong decision．n啦complexky of the system is reduced and the recognition ofthe system is improved． Key