摘要
针对单一的人脸特征在识别中的局限性,提出了一种基于特征融合的人脸识别方法,首先利用主成分分析获得原始输入图像的特征脸,经图像重构处理得到原始图像的余像,然后抽取余像的特征脸,最后将两种特征脸按一定的权重融合成一个组合特征进行人脸识别,通过针对ORL人脸数据库的实验表明:该特征融合方法的人脸识别是行之有效的,优于传统特征脸的方法,识别率可以达到91.5%.
A new face recognition method based on features fusion is suggested in order to overcome the limitation of sole facial feature in recognition.Firstly,the eigenface of original face images is extracted by the technology of principal component analysis,after the image is obtained by reconstructing face images.Secondly,the second order eigenface is obtained in the same way.Finally these two kinds of eigenface are fused into a combined one over their respective matching score weighted by a set of coefficients.The method is evaluated on ORL database based on the third neighbor classifier.The experimental results show that eigenface fusion method for face recognition have high feasibility,and the recognition rate is better than traditional eigenface method,with the accuracy rate reaching 91.5%.
出处
《湖北民族学院学报（自然科学版）》
CAS
2011年第2期188-190,共3页
Journal of Hubei Minzu University(Natural Science Edition)
基金
湖北省自然科学基金项目(2009CDB069)
恩施州科技局项目"基于ICA的人脸识别技术在门禁系统的应用研究"
湖北民族学院青年项目"嵌入式网络视频监控系统研究"
关键词
特征脸
人脸识别
特征融合
主成分分析
eigenface
face recognition
features fusion
principal component analysis
