基于单演特征和协同表示的人脸识别方法

基于协同表示的分类方法解决了稀疏表示分类方法太过强调l1模的问题被广泛应用于人脸识别中。为了进一步提高鲁棒性和识别率,提出了基于单演特征的协同表示分类方法,即MCRC。单演特征所提取的图像相位信息对光照的鲁棒性强并且其方向信息和幅值信息对姿态的鲁棒性也很高,相对于Gabor特征的多尺度和多方向,单演特征在特征变换的速度上也具有一定优势。在AR、LFW人脸数据库上的实验结果表明,该方法具有可行性和有效性。

FACE RECOGNITION BASED ON MONOGENIC FEATURES AND COLLABORATIVE EPRESENTATION

The collaborative representation based classification has solved the problem of the sparse representation based classification, which is too much emphasis on l₁ norm is widely used in face recognition. In order to further improve the robustness and recognition rate, we propose a face recognition method based on monogenic features and collaborative representation, namely MCRC. The robustness of the image phase information extracted by the monogenic feature is robust to the illumination and the robustness of the direction information and the amplitude information to the attitude is also high. Compared with the multi-scale and multi-direction of the Gabor feature, the speed of the transformation also has certain advantages. The experimental data on the AR, LFW face database show that the method is feasible and effective.