改进的分块2DPCA人脸识别方法

将样本中间值融入模块二维主成分分析方法进行人脸识别。该算法首先对图像矩阵进行了模块化得到子图像矩阵,之后直接把子图像矩阵集作为样本集。与原始模块二维主成分分析算法不同之处在于,将子块的类内中间值引入到总体协方差矩阵的求解过程中。通过ORL数据库的测试,融合后的算法继承了模块二维主成分分析的强鲁棒性,提高了识别率,证明了改进后的方法相对普通的二维主成分分析和模块二维主成分分析算法而言,性能得到提升。

Face Recognition Method based on Improved Modular 2DPCA

The within-class median is integrated into Modular 2DPCA method.First the image is divided into modular sub-images.Then,the sub-images obtained from upper state are directly taken as the sample set.Compared to the original modular 2DPCA,the within-class median of the sub-images is introduce into the solution-finding process of the general matrix.The experiment results based on ORL face database indicate that the proposed method is better and more robust than traditional 2DPCA and modular 2DPCA in recognitio...