改进的主成分分析和最近邻的人脸识别方法

The Improved PCA and NNA Method of Face Recognition

在传统的主成分分析特征提取中,直接求解特征空间是很困难的,同时也是非常浪费资源,为优化这一问题,该文提出了改进的主成分分析特征提取。在人脸特征提取中,同时还选择了适当的主分量数,用于提高分类识别的速度。在人脸分类识别的过程中,分类策略选取最邻近分类器,通过计算最短欧几里得距离来分类识别测试样本。通过十折交叉验证方法验证了改进的主成分分析和最邻近分类的有效性。

In the traditional principal component analysis（PCA） feature extraction,directly solving feature space is very difficult and very wasteful.For optimizing this problem,an improved PCA feature extraction method is proposed.In the process of face feature extraction,at the same time,the proper principal components number is selected to enhance the speed of classification.In the process of face classification,the classification strategy is the nearest neighbor algorithm（NNA）,through calculating the shortest distance to classify and recognize the test samples.The ten-fold cross-validation method proves the validity of the improved PCA and the NNA.