计算机工程与应用 ›› 2013, Vol. 49 ›› Issue (5): 166-169.

• 图形图像处理 • 上一篇    下一篇

LBP与LNMF特征融合的人脸识别

袁宝华1,王  欢2,任明武2   

  1. 1.南京理工大学 泰州科技学院 计算机系,江苏 泰州 225300
    2.南京理工大学 计算机系,南京 210094
  • 出版日期:2013-03-01 发布日期:2013-03-14

Fusing local binary pattern and LNMF of face recognition

YUAN Baohua1, WANG Huan2, REN Mingwu2   

  1. 1.Department of Computer Science & Technology, Taizhou Institute of Science & Technology, Nanjing University of Science and Technology, Taizhou, Jiangsu 225300, China
    2.School of Computer Science & Technology, Nanjing University of Science and Technology, Nanjing 210094, China
  • Online:2013-03-01 Published:2013-03-14

摘要: 提出一种融合局部二值模式(LBP)和局部非负矩阵分解(LNMF)进行人脸识别的方法,采用LBP算子提取分块人脸图像的LBP直方图序列(LBPHS),根据每块的贡献度,得到权重的直方图序列(Weight LBPHS),采用LNMF方法提取其非负子空间及其系数矩阵,根据最近邻原则进行识别。在ORL和YALE标准人脸数据库上的实验表明,该方法具有较高的识别率。

关键词: 局部二值模式, 局部非负矩阵分解, 人脸识别

Abstract: A method of face recognition based on Local Binary Pattern(LBP) and Local Non-negative Matrix Factorization(LNMF) is proposed. LBP operator is used to extract the LBP Histogram Sequence(LBPHS) from block face images. According to the contribution of each face block, weight LBP Histogram Sequence(Weight LBPHS) is obtained. LNMF is applied to weight LBPHS for extracting non-negative subspace and the corresponding coefficient matrices. Nearest neighbor principle is utilized in face recognition. The simulation experiments illustrate that this method has better recognition rate on the ORL and YALE face database.

Key words: local binary pattern, local non-negative matrix factorization, face recognition