基于双向主成分分析和压缩感知的人脸识别算法

Face Recognition Algorithm Based on Bidirectional Principal Component Analysis and Compressed Sensing

提出了一种双向主成分分析(BD-PCA)与基于光滑l0范数(SL0)相结合的人脸识别算法(BP-SL0)。首先利用BD-PCA对所有的训练图像降维,然后将降维后的人脸图像按列拉伸成一个向量,并将其组成字典矩阵,同时对待测试图像进行相同处理,最终通过SL0算法求解优化问题。实验结果表明,该算法获得了较高的识别率和重建效果,且效果优于单独使用BD-PCA和SL0算法。

This paper proposes a face recognition algorithm combining the bidirectional principal component analysis （BD-PCA） and the face recognition algorithm based on smooth l0 norm（SL0）. First we use BD-PCA to reduce the dimension of these training images, and then make each face image into a vector by stretching it according to each column, and use these vectors to comprise a dictionary matrix, and process test images in the same way, and finally solve the optimization problem by SIX） algorithm. Experimental results show that the proposed algorithm achieves high recognition rate and better reconstruction effect than those by BD-PCA and SLO algorithm alone.