摘要
针对曲线特征更能反映人脸图像的主要特征和独立成分分析能够提取高阶信息的优势,提出了一种基于曲波变换与独立成分分析的人脸识别方法.首先将人脸图像进行曲波变换,选择粗尺度层系数作为曲波特征,然后对曲波特征下采样后进行独立成分分析,提取部分独立成分构成特征空间,最后根据最近邻分类器分类.在0RL和Yale人脸库上的相关实验表明:该方法在识别性能方面优于对比方法.
As the main features of the faces can be better represented by the curvelet coefficients, and higher-order feature can be extracted by independent component analysis, a method of face recognition based on curvelet transform and ICA is proposed in this paper. Firstly, each of the images is decomposed using curvelet trasnform, and the low-frequency face image is selected as a sub-image;secondly,ICA is adopted to obtain independent components, and part of independent components are selected to constitute the feature space. Finally, the nearest neighbor classifier is used for identification. The experiment result on ORL and Yale face databases shows that the proposed method improved the recognition performance in comparison with comparative approach.
出处
《湖北民族学院学报（自然科学版）》
CAS
2015年第3期300-303,共4页
Journal of Hubei Minzu University(Natural Science Edition)
基金
湖北省自然科学基金项目(2013CFB042)
关键词
人脸识别
曲波变换
独立成分分析
特征提取
最近邻分类器
face recognition
curvelet transform
independent component analysis
feature extract
the nearest neighbor classifier
