摘要
提出了一种新的小波域主元分析与线性辨别分析相结合的红外人脸识别方法。首先通过DWT将红外人脸图像通过二级小波分解成七个子带,舍去两次分解中的对角子带,对剩下的五个子带进行有效的组合;然后用PCA方法对组合后的向量进行特征提取,再把PCA提取的特征向量进行线性辨别分析;最后用欧氏距离和三近邻分类器得到分类结果。同传统的PCA和PCA+LDA的方法相比,该方法更能利用人脸图像的有用判别信息,并得到更好的识别效果。
A novel method for infrared face recognition based on PCA and LDA in wavelet domain is proposed, Firstly, each infrared face image was decomposed into seven sub-bands using tow scales＇ DWT, Secondly the sub-bands were arranged for PCA approach to extract the features. Then, use LDA approach to further features extraction, Finally, Euclidean distance and the 3-NN classifier were utilized in recognition. The experiments illustrate that the proposed method has better performance compared with traditional PCA and PCA ＋ LDA.
出处
《计算机应用研究》
CSCD
北大核心
2008年第5期1586-1588,共3页
Application Research of Computers
基金
国家自然科学基金资助项目(60665001
60662003
60462003
10626029)
江西省自然科学基金资助项目(0611025)
关键词
离散小波变换
主元分析
线性辨别分析
红外人脸识别
discrete wavelet transform
principal component analysis
linear discriminant analysis
infrared face recognition
