分块CS-LBP和加权PCA的低分辨率人脸识别

Low-resolution face recognition based on blocking CS-LBP and weighted PCA algorithm

针对局部二值模式(LBP)特征在低分辨率的人脸图像上识别率较低的问题,提出了一种基于分块中心对称局部二值模式(CS-LBP,center symmetric local binary pattern)和加权主成分分析(PCA)算法的低分辨率人脸识别算法。首先利用分块CS-LBP算子提取低分辨率人脸图像的特征;然后利用加权PCA算子对特征进行降维,从而得到更强的分类特征;最后利用最近邻分类器选出人脸最优分类类别并计算识别率。在ORL人脸库上的实验表明,在人脸图像分辨率下降到(12×10)时,本文算法的识别率仍能达到85.00%,基本满足了实际运用中对识别率的要求,并且降低了运算时间。

To improve the recognition accuracy of Local Binary Pattern （LBP） on low-resolution face rec- ognition. A novel method is proposed in this paper by combining blocking center symmetric local binary pattern （CS-LBP） and weighted principal component analysis （PCA）. Firstly, the features of low-resolu tion/ace images are extracted by blocking CS-LBP operator. Secondly, the stronger classification and lower dimension features can be got by applying weighted PCA algorithm. Finally the distance is calcu- lated and used to select the optimal classification categories of low-resolution face set by using the nea- rest neighbor classifier. Besides, the recognition rate also can be calculated. The experimental results in the ORL human face database show that recognition rate can reach 85.00 % when resolution of face im- age drops to （12 × 10）, which can basically satisfy the practical requirements of recognition and reduce computation time on low-resolution face recognition.