基于MB-LBP算子和Multilinear PCA算法的人脸识别

针对运用MB-LBP算法提取的人脸特征维数较高、而直接用MB-LBP算法提取的特征进行人脸识别时计算量较大的问题,提出一种融合MB-LBP和Multilinear PCA算法的新的人脸识别方法。首先利用MB-LBP算法提取人脸图像的特征;然后用Multilinear PCA算法对提取的人脸特征进行降维;最后用最近邻分类器进行人脸识别。在FERET人脸库上进行验证,实验结果表明,该方法的识别率高于传统PCA、分块PCA、LBP和PCA相结合的方法。

Face recognition based on MB-LBP operator and multi-linear PCA algorithm

The dimension of facial features extracted by utilizing the MB-LBP algorithm is too high, as a result it increases the computation greatly when utilizing the MB-LBP algorithm in face recognition. This paper proposed a hybrid method combined MB-LBP algorithm with multilinear PCA algorithm. First, it utilized the MB-LBP algorithm to extract facial images. Then it utilized the multilinear PCA algorithm to reduce the dimensionality of the extracted features. Finally, it utilized the nearest neighbor classifier for face recognition. It conducted an experiment, and prored the hybrid algorithm on the FERET face database. Experiment results show that the recognition rate of the algorithm utilized is higher than the traditional PCA, muti-block PCA or combining them together.