Abstract: To overcome the limitation of traditional single sample face recognition, a new method of face recognition based on Monogenic features and CS-LBP（MCSLBP） is proposed. Center-Symmetric Local Binary Pattern（CS-LBP） is firstly adopted to encode the monogenic magnitude on the same scale. The monogenic phase is quantified into four regions while encoded in horizontal direction and vertical direction. Then these three parts are integrated into MCSLBP feature. Finally, MCSLBP feature map at different scales is divided into several blocks, and the concatenated histogram features calculated over all blocks are used for the feature descriptor of face recognition, and the recognition is performed by using the nearest neighbor classifier. Experimental results on CAS-PEAL and AR databases show that the MCSLBP algorithm is an outstanding method for single sample face recognition under different illumination conditions, different facial expression conditions and partial occlusion conditions.

Key words: face recognition, single sample, monogenic signal, center-symmetric, magnitude phase and orientation pattern, Center-Symmetric Local Binary Pattern（CS-LBP）

摘要： 针对单样本情况下传统人脸识别方法识别效果不佳的问题，提出一种融合单演幅值、相位和方向的单演中心对称幅值相位方向模式（MCSLBP）的人脸识别方法。首先采用中心对称局部二值模式（CS-LBP）对同一尺度下的单演幅值进行编码，并将单演相位量化到4个区间进行编码，同时对单演水平方向和垂直方向进行二值编码，然后将三者融合成MCSLBP特征；最后对不同单演尺度空间中的MCSLBP模式图进行分块，提取每一小块的直方图特征并串联后用最近邻分类器进行分类识别。在CAS-PEAL和AR人脸库上的实验结果表明，MCSLBP方法对具有光照、表情和遮挡变化的单样本人脸识别具有较好的识别效果。

关键词: 人脸识别, 单样本, 单演信号, 中心对称, 幅值相位方向模式, 中心对称局部二值模式（CS-LBP）