基于小波变换和多分类器的人脸识别方法  

FACE RECOGNITION BASED ON WAVELET TRANSFORM AND MULTIPLE CLASSIFIER

作　　者：苏宏涛[1] 赵荣椿[1] 等 

机构地区：[1]西北工业大学计算机科学与工程系，陕西西安710072

出　　处：《计算机应用》2002年第8期25-27,共3页journal of Computer Applications

摘　　要：现在已有许多特征用于人脸识别 ,而不同的特征反映了图像的不同特性。因此 ,结合多个特征 ,使用多分类器来进行分类可以提高识别率。文中在对原始图像进行小波变换预处理的基础上 ,抽取本征脸特征和奇异值特征 ,并利用对应着两类特征的多分类器进行分类。利用ORL人脸库进行了实验 。Many features have been used in face recognition, and different features reflect the different property of the image, so it will improve the ability of recognition by using combining multiple features and multiple classifier. After the pre-processing of wavelet transform, the low frequency image was regarded as the input to be processed in next steps. The eigenface feature and singular value feature were extracted, and the classifying was implemented by using the different classifiers corresponding to the different features and a combined classifier corresponding to all features extracted. The ORL face database was used in the experiment and the result show that the method presented is effective.

关 键 词：小波变换 多分类器 人脸识别 奇异值特征 图像预处理 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]