二维最大散度差图像投影鉴别分析  

Two-dimensional Maximum Scatter Difference Image Projection Discriminant Analysis

作　　者：陈才扣[1] 刘永俊[2] 杨静宇[1] 

机构地区：[1]南京理工大学计算机科学与技术学院,江苏南京210094 [2]扬州大学信息工程学院,江苏扬州225009

出　　处：《系统仿真学报》2007年第4期833-835,913,共4页Journal of System Simulation

基　　金：国家自然科学基金(60472060);江苏省博士后科研资助计划项目;江苏省高校自然科学基金(05KJB520152)

摘　　要：提出了一种新的二维散度差图像投影鉴别分析方法。该方法利用类间离散度与类内离散度之差作为鉴别准则,从根本上避免了传统的Fisher线性鉴别分析所遇到的小样本问题时。所提出的方法是直接基于图像矩阵的,与以往的基于图像向量的鉴别方法相比,它的突出优点是大大提高了特征抽取的速度。在ORL人脸数据库和AR标准人脸库上的仿真试验结果表明,所提出的方法不仅在识别性能上优于传统的散度差鉴别分析,特征抽取的速度有了较大幅度的提高。A novel image projection discriminant analysis based on scatter difference criterion was developed for image feature extraction. The proposed one adopts the difference of both between-class scatter and within-class scatter as discriminant criterion, In such a way, the small sample size problem occurred in traditional Fisher discriminant analysis is in nature avoided. In addition, the construction of scatter matrices is directly based on original training image matrices rather than vectors. It is not necessary to convert the image matrix into high-dimensional image vector like those previous linear discriminant methods so that much computational time would be saved if using the proposed method for feature extraction. Finally, extensive experiments were performed on ORL face database and AR face database. The experimental results indicate that the proposed method outperforms the traditional scatter difference discriminant analysis in recognition performance. And, the speed for feature extraction is greatly improved.

关 键 词：散度差鉴别准则 图像矩阵 图像投影鉴别分析 人脸识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]