摘要
提出了一种基于分块小波变换与奇异值阈值压缩的人脸特征提取与识别算法。该方法首先对人脸图像进行分块小波变换,并根据图像块的位置分布选取不同的频率分量,然后对该分量进行奇异值阈值压缩与特征融合,最后在ORL人脸库上利用最近邻分类器对该特征进行分类识别,验证了算法的有效性。
A feature extraction and recognition algorithm of intersected human face based on wavelet transform and SVD （ Singular Value Decomposition） threshold compression is proposed. Firstly, the intersected human faces are transformed with wavelet, and different wavelet coefficients are extracted according to the position of intersected images. With SVD threshold compression, the coefficients are compressed and fused into ultimate discriminate features. Finally, the features are classified and recognized by the Nearest Neighbor Classifier on ORL face database. The experimental results validate the efficiency of the algorithm.
出处
《计算机应用与软件》
CSCD
北大核心
2008年第1期30-32,78,共4页
Computer Applications and Software
基金
国家自然科学基金项目(60472060,60473039)资助。
关键词
人睑分块
小波变换
奇异值闽值压缩
特征融合
Intersected human face Wavelet transform SVD threshold compression Feature fusion
