计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (19): 146-149.DOI: 10.3778/j.issn.1002-8331.2009.19.045

• 图形、图像、模式识别 • 上一篇    下一篇

基于虚拟信息的单样本分块人脸识别

吴 朋1,周激流1,李晓华1,丁晓莲2   

  1. 1.四川大学 计算机(软件)学院,成都 610065
    2.四川大学 电子信息学院,成都 610065
  • 收稿日期:2008-04-17 修回日期:2008-09-01 出版日期:2009-07-01 发布日期:2009-07-01
  • 通讯作者: 吴 朋

Sub-block face recognition based on virtual information with one training image per person

WU Peng1,ZHOU Ji-liu1,LI Xiao-hua1,DING Xiao-lian2   

  1. 1.College of Computer Science,Sichuan University,Chengdu 610065,China
    2.School of Electronics and Information Engineering,Sichuan University,Chengdu 610065,China
  • Received:2008-04-17 Revised:2008-09-01 Online:2009-07-01 Published:2009-07-01
  • Contact: WU Peng

摘要: 在单样本人脸识别中,由于训练样本的数量受限,充分利用样本的信息就显得十分重要;针对这种情况,通过采用旋转和反转原始图像的方法增加样本信息,提出了基于虚拟信息的单样本分块人脸识别方法,充分利用了样本的整体信息和局部信息。实验表明,在对人脸图像进行识别时取得了较好的效果,在一定程度上克服了单样本下姿态对识别效果的强烈影响。

关键词: 虚拟样本, 特征提取, 分块, 图像旋转

Abstract: As the number of training samples limited in the face recognition with one training sample per person,it is quite important to make full use of the information of the samples.Aiming at this situation,a method of sub-block face recognition based on virtual information with one training image per person is proposed,which makes the best of the globe and local information of the samples.Compared by the examples,the proposed method attains a better result when recognizes face image.At a certain extent,it can overcomes the intensive influence of pose to the recognition effect.

Key words: virtual information, feature extraction, sub-block, image circumgyration