典型相关分析融合全局和局部特征的人脸识别

为了提高人脸的识别率,提出一种典型相关分析融合全局和局部特征的人脸识别算法(SUB-CCA)。通过划分子模式方式避免人脸识别存在小样本、非线性问题,并提取局部特征,采用主成分分析提取人脸图像的全局特征,并采用相关分析算法对全局、局总特征进行融合,消除特征间冗余信息,降低特征维数,采用投票法得到人脸识别结果,并采用3个人脸数据集对算法性能进行测试。仿真结果表明,相对于参比算法,SUB-CCA提高了人脸识别的识别精度。

Face automatic recognition algorithm based on canonical correlation analysis fusion global and local features

In order to improve the recognition rate of face image, a novel face recognition method is proposed based on sub-pattern and canonical correlation analysis. The global and local features are extracted, and the redundant information between the features is eliminated, and then the face images are divided in sub models to avoid small sample, nonlinear problems, and the recognition results are corrected by voting method to increase stability of the algorithm, three face data sets are used to test the performance of the algorithm. The simulation results show that, SUB-CCA improves the recogni-tion rate of face image compared with other algorithms.