基于动态优化PCA人脸识别算法的实现

提出一种人脸识别动态优化PCA算法。采用ORL人脸数据库，对数据库中图像进行放缩归一化操作并按升序或降序排列。然后，将数据库分为最大值类、中值类、最小值类三个部分，求其各类平均值，特征值个数选取9和10，实现人脸识别和重建。随机采样人脸图像，采用DCT算法将其转换到频域进行分析，通过比较整个数据库平均脸与三类中各自平均脸，算法运行后，特征值个数可以实现自动优化。实验结果显示，该算法在一定程度上可用来对传统PCA算法部分关键参数进行优化。

Face Recognition Based on Dynamically Optimized PCA Method

This paper addresses an algorithm for dynamically optimizing PCA method in face recognition. Firstly,all the resized human face images in the ORL database are performed with DCT algorithm and then rearranged by an increasing or ascending order. Secondly, the data are divided into three groups compared with the conventional PCA method,i. e. the Maximum,the Middle and the Minimum. The mean average images are also calculated for such three groups. In PCA process,the Eigen value number is selected as nine and the typi-cal ten. Finally,recognition and reconstruction may be completed. In testing period,some random face images are selected and also con-verted into frequency domain by DCT. By fitting the samples mean average and those three groups＇ones,the optimization is selected. The Eigen value number will be dynamically optimized when the algorithm runs. The experimental results show that this method has a better capability and can be used to realize dynamically optimizing the key parameter of PCA and achieve an acceptable result to some extent.