作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2006, Vol. 32 ›› Issue (14): 179-180,. doi: 10.3969/j.issn.1000-3428.2006.14.066

• 人工智能与识别技术 • 上一篇    下一篇

模块二维主成分分析——人脸识别新方法

陈伏兵1, 2,陈秀宏1, 2,张生亮2,杨静宇2   

  1. (1. 淮阴师范学院数学系,淮安 223001;2. 南京理工大学计算机科学系,南京 210094)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2006-07-20 发布日期:2006-07-20

Modular Two Dimensional Principal Component Analysis
——A Novel Method for Human Face Recognition

CHEN Fubing1, 2, CHEN Xiuhong1, 2, ZHANG Shengliang2, YANG Jingyu2   

  1. (1. Department of Mathematics, Huaiyin Teachers College, Huaian 223001;
    2. Department of Computer Science, Nanjing University of Science and Technology, Nanjing 210094)
  • Received:1900-01-01 Revised:1900-01-01 Online:2006-07-20 Published:2006-07-20

摘要: 提出了模块二维主成分分析(M2DPCA)线性鉴别分析方法。M2DPCA方法先对图像矩阵进行分块,对分块得到的子图像矩阵直接进行鉴别分析。其特点是:能有效地降低模式原始特征的维数;可以完全避免使用矩阵的奇异值分解,特征抽取方便;此外,2DPCA是M2DPCA的特例。在ORL人脸库上试验结果表明,M2DPCA方法在识别性能上优于PCA,比2DPCA更具有鲁棒性。

关键词: 线性鉴别分析, 特征抽取, 特征矩阵, 人脸识别

Abstract: This paper presents modular two dimensional principal component analysis (M2DPCA)——a novel technique for human face recognition. First, the original images are divided into sub-images images in proposed approach. Then, the well-known 2DPCA method can be directly used to the sub-images obtained from the previous step. There are two advantages for this way: (1)dimension reduction of original pattern features can be done efficiently; (2)singular value decomposition of matrix is absolutely avoided in the process of feature extraction so the discriminant features can be gained easily. Moreover, 2DPCA is the special case of M2DPCA. To test and to evaluate the performance of M2DPCA, a series of experiments are performed on ORL human face image database. The experimental results indicate that the recognition performance of M2DPCA is superior to that of PCA and is robust than that of 2DPCA at the same time.

Key words: Linear discriminant analysis, Feature extraction, Feature matrix, Face recognition

中图分类号: