一种改进的基于二维局部保留投影的人脸识别方法  

Face Recognition Based on Improved 2DLPP Method

作　　者：郑钢[1] 杨丹[1] 张小洪[1] 方莉莉[1] 

机构地区：[1]重庆大学软件学院,重庆400030

出　　处：《重庆工学院学报（自然科学版）》2008年第11期142-145,共4页Journal of Chongqing Institute of Technology

基　　金：重庆市自然科学基金项目(CSTC2005BA2002);重庆大学研究生科技创新基金(200801A1B0290288)

摘　　要：提出了一种改进的基于二维局部保留投影的人脸识别方法.首先使用2DPCA进行降维处理,最大限度保留显著的二维信息,在降维后的特征空间里,使用优化2DLPP方法构建出最优投影矩阵,优化的2DLPP能够保留更多的特征矩阵内在的流行结构,相比将特征矩阵作为一个整体的2DLPP方法具有更好的识别率.由于同时在行、列方向上应用了2DPCA和2DLPP方法,使得在图像的表示和识别过程中产生了较少的参数,提高了整个算法的运行效率.最后用欧式距离和最近邻分类器实现了分类识别.通过在ORL人脸库的对比实验,改进的算法具有更高的识别准确率和速度.This paper presents a method for face recognition based on an improved 2DLPP. 2DPCA is first utilized for dimensionality reduction of feature space maintaining most prominent 2D information. Thus similarity matrix corresponding to elements is easily constructed and the followed 2DLPP can be implemented directly in the reduced feature space. The improved method with more intrinsic manifold structure of feature matrices yields higher recognition accuracy than the existing 2DLPP which treats the feature matrices as a whole. Meanwhile, fewer coefficients are extracted for image representation and recognition owing to 2DLPP and 2DPCA in the row and column directions simultaneously. Euclidean distance and the nearest classifier are finally used for classification. Experiment results demonstrate the effectiveness of the proposed method in both recognition accuracy and speed.

关 键 词：2DLPP 2DPCA 人脸识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]