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 ›› 2011, Vol. 37 ›› Issue (2): 175-177. doi: 10.3969/j.issn.1000-3428.2011.02.060

• 人工智能及识别技术 • 上一篇    下一篇

局部嵌入投影及其在人脸识别中的应用

喻以明,陈才扣   

  1. (扬州大学信息工程学院,江苏 扬州 225009)
  • 出版日期:2011-01-20 发布日期:2011-01-25
  • 作者简介:喻以明(1985-),男,硕士研究生,主研方向:人脸识别,人工智能;陈才扣,副教授、博士
  • 基金资助:
    国家自然科学基金资助项目(60875004)

Locality Embedding Projection and Its Application to Face Recognition

YU Yi-ming, CHEN Cai-kou   

  1. (Information Engineering College, Yangzhou University, Yangzhou 225009, China)
  • Online:2011-01-20 Published:2011-01-25

摘要: 针对线性鉴别分析忽略样本间局部结构特性的缺陷,提出一种局部嵌入投影人脸识别算法。利用样本间的近邻关系和类别标签信息将所有样本分属于多个近邻类和非近邻类;采用局部均值思想,对每个样本所对应的近邻类和非近邻类,定义其类内散布和类间散布;通过最大化总体类间散布与总体类内散布之比,使得具有相同类别标签且互相邻近的样本点在投影空间中尽可能靠近,而相互邻近的类彼此远离。ORL人脸库和FERET人脸库上的实验结果证实了算法的有效性。

关键词: 局部均值, 特征抽取, 人脸识别

Abstract: A main defect of Linear Discriminant Analysis(LDA) is that local geometric features are ignored. To address this problem, a novel manifold-based method named Locality Embedding Projection(LEP) algorithm is presented in this paper. In the algorithm, neighborhood relationship and class label information are used to classify training sample set. For each training sample, there are two classes which are called neighbor class and non-neighbor class; then inter-class scatter and intra-class scatter are defined for each training sample. The ratio of total inter-class scatter and total intra-class scatter is maximized to make nearby samples with the same class-label are more compact, and nearby classes are separated. Experimental results on ORL and FERET face databases show effectiveness of the proposed method.

Key words: local mean, feature extraction, face recognition

中图分类号: