HOG和多流形判别分析相融合的人脸识别

Face recognition based on HOG and discriminative multi-manifold analysis

为了提高人脸的识别率,提出一种方向梯度直方图和多流形判别分析相融合的人脸识别算法。将一幅图像划分为多个子块图像块,并采用方向梯度直方图算法对每一个小块进行处理;采用多流形判别分析提取特征,并采用最小二乘支持向量机建立分类器对人脸进行分类和识别;在Yale和AR人脸库进行仿真实验。实验结果表明,相对于传统人脸别算法,该算法不仅提高了人脸识别率和识别速度,并且对光照和姿态变化具有较强的鲁棒性。

In order to improve the recognition rate of face,a new face recognition method is proposed based on histogram oriented gradient and discriminative multi-manifold.Firstly,the face image is divided into several patches and the patches are processed by histogram oriented gradient to form an image set.Secondly,discriminative multi-manifold is used to select the features,and least squares support vector machine is used to establish the classifier to recognized the face.Finally,the simulation experiments are carried out to test the performance on Yale and AR face data set.The experimental results show that the proposed method can improve the recognition rate and recognition speed compared with traditional face recognition methods,and has well robust for face recognition in illumination and pose conditions.