基于小波变换与小域特征模糊融合的人脸识别  

Face recognition based on wavelet transform and small-area feature fuzzy fusion

作　　者：郑德忠[1] 崔法毅[1] 

机构地区：[1]燕山大学电气工程学院,河北秦皇岛066004

出　　处：《光学技术》2008年第6期841-846,共6页Optical Technique

基　　金：国家教育部博士点基金资助项目(2006021600)

摘　　要：小波变换是一种很好的图像压缩方法,利用小波变换对人脸图像进行三次小波分解,并将低频分量分割成为7个子图像。鉴于人脸上的各小域子图像信息的相互独立性。先利用小域子图像实现软分类,然后使用传统奇异值分解(SVD)法提取出各小域子图像的奇异值(SV),构造出小域奇异值特征向量,给出待识别图像对训练样本图像的隶属度,并采用模糊融合的方法对小域特征进行数据融合,获得识别结果。实验结果表明,该方法实现起来简单、识别速度快,具有很高的识别率。Wavelet transform is a very good method about image compression. Human face images are decomposed using wavelet transform, and parts of low frequency are divided into seven sub-images. Owing to information independence of facial small-area sub-images, soft recognition based on small-area sub-images can be implemented firstly. The traditional SVD （singular value decomposition） operates directly on a set of new training sub-images respectively. A set of small-area SV can be extracted and then a set of small-area SV eigenvector can be obtained. According to these small-area sub-features, the membership grades of the test sub-images to the training sub-images can be determined. The identity of an unknown face image is determined by the fuzzy fusion which aggregates the small-area sub-features. The experiments show that the proposed method has the characteristies of simple realization, rapid recognition speed and high recognition rate.

关 键 词：小波变换 小域奇异值特征 模糊融合 自适应加权 隶属度 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]