摘要： 提出了一种融合小波矩描述子(WMD)矩阵与二维主成分分析(2DPCA)的人脸特征抽取与识别算法。该方法抽取描述人脸本质特征的WMD矩阵，利用2DPCA对该矩阵进行投影压缩降维，抽取人脸最终鉴别特征，利用最近邻分类器对特征进行分类识别。NUST603人脸库上的实验结果验证了算法的有效性。
关键词:
小波矩描述子,
二维主成分分析,
人脸,
特征抽取
Abstract: This paper proposes a feature extraction and recognition method of human face based on wavelet-moment descriptors and 2-dimensional principal component analysis(2DPCA). With this method, the matrix of wavelet-moment descriptors which describes the human face image’s essential feature is extracted, the matrix is projected and compressed with 2DPCA and the ultimate discriminant features is obtained. The features are classified with the nearest neighbor classifier. Experimental results on NUST603 face database confirm the efficiency of this method.
Key words:
wavelet-moment descriptors,
2-dimensional principal component analysis(2DPCA),
human face,
feature extraction
中图分类号:
谢永华;陈伏兵;张生亮;杨静宇. 融合WMD矩阵与2DPCA的人脸特征抽取与识别[J]. 计算机工程, 2007, 33(16): 23-25.
XIE Yong-hua; CHEN Fu-bing; ZHANG Sheng-liang; YANG Jing-yu. Human Face Feature Extraction and Recognition Based on Wavelet-moment Descriptors and 2DPCA[J]. Computer Engineering, 2007, 33(16): 23-25.