摘要
提出一种基于Contourlet变换域的稀疏表示分类方法对人脸进行识别,使用Contourlet波变换对初始图像进行处理,得到原始图像的低频和高频特征,将低频分量与高频分量直接组合为一维向量,输入稀疏表示分类算法进行识别.研究结果表明,该方法能够对图像进行快速特征提取,去除噪声和冗余,保留边缘等局部特征,同时降低了图像维数.与PCA+SRC、LPP+SRC方法相比,该方法能够得到更好的判别特征和更高的识别率.
A kind method based on sparse representation and the Contourlet transform for face recognition is proposed.The original images are filtered by the Contourlet transform,the low and high frequency characteristics are obtained.Then combining these two kinds of characteristics into one dimensional vector,the vector is put into the sparse coding algorithm for the face recognition.The results shows that the proposed method can process the image feature extraction rapidly,remove noise and redundancy,retain local image features,and reduce the data dimensionality.The proposed method can obtain better features and better face recognition performance by comparison with PAC＋SRC method and LPP＋SRC method.
出处
《内蒙古师范大学学报（自然科学汉文版）》
CAS
北大核心
2016年第1期89-93,共5页
Journal of Inner Mongolia Normal University(Natural Science Edition)
基金
国家自然科学基金资助项目(61201370)
山东省自然科学基金资助项目(ZR2014FM039)
关键词
稀疏表示分类方法
CONTOURLET变换
低频子带
高频子带
人脸识别
sparse representation for classification
contourlet transform
low frequency characteris tics
high frequency characteristics
face recognition
