摘要
人脸识别是生物特征识别和人工智能领域特别重要的课题之一。讨论了统计主成分分析法(Principal Compo-nent Analysis,PCA)在人脸识别中的应用。PCA是基于统计的方法,可以对人脸库数据起到降低维数、去除相关性等作用。通过Kauhunen-Loeve变换(K-L变换)将人脸库变换到新的坐标系,得到人脸特征子空间,然后将待测人脸图像投影到特征子空间,最后利用2-范数距离分类器进行分类,从而达到识别的目的。最后利用人脸库对其进行了测试。
Face recognition is an especially important subject in the area of biometrical recognition technology and artificial intelligence. In the paper, the application of Principal Component Analysis in the face recognition is discussed. PCA is based on the statistics, and can play an role of reducing the dimension and dependent of the face image database data. The feature subspace of face can be obtained after transformed the image face database to a new coordinate system through Kauhunen- Loeve （K-L） transformation. Then project the test image to feature subspace; At last, the projection result is classified using the 2-norm distance classifier to achieve the goal of recognition. The test of face image database with PCA is presented.
出处
《电子设计工程》
2011年第20期101-104,109,共5页
Electronic Design Engineering
