ISSN 1000-1239 CN 11-1777/TP

计算机研究与发展 ›› 2014, Vol. 51 ›› Issue (10): 2302-2307.doi: 10.7544/issn1000-1239.2014.20130822

• 人工智能 • 上一篇    下一篇



  1. (北方工业大学信息工程学院 北京 100144) (
  • 出版日期: 2014-10-01
  • 基金资助: 

Dynamic Spatial Synthesis of Dorsal Hand Vein Images Based on PCA

Wang Yiding, Jiang Nan, Li Kefeng   

  1. (College of Information Engineering, North China University of Technology, Beijing 100144)
  • Online: 2014-10-01

摘要: 目前对手背静脉识别问题的研究大多是在较小规模的数据上进行,几乎没有在大样本情况下对手背静脉识别进行实验.因此,为了扩充手背静脉样本库提出了一种新的手背静脉图像合成方法,其基本思想是源于PCA(principal component analysis)原理,将用于合成的样本分为2组,对一组进行主成份分析构造特征空间,再由另一组向特征空间投影得到的投影系数构造投影空间,最后利用投影空间的投影系数在特征空间上进行PCA重建,从而融合双空间的信息达到图像合成的目的.通过对分组选取的动态更新,可以大量地合成手背静脉图像样本.由此,在实际实验中在拥有94个人的原始图像数据库的基础上建立一个拥有8007个人的合成图像数据库.合成图像数据库的识别率达到97.84%.良好的识别率说明了合成图像数据库今后可以用于手背静脉相关的模拟测试中.

关键词: 主分量分析, 手背静脉, 图像合成, 动态空间, 生物识别

Abstract: Databases of human dorsal hand vein (DHV) images are created and distributed for the purpose of testing DHV identification algorithm. For logistical and privacy reasons, these databases are often too small to fulfill their potential applications. A novel synthesis method of creating new DHV images using principal component analysis (PCA) is proposed, which can be applied to enlarge the existing DHV image database. Firstly, the origin database is divided into set B and set M. Set B provides a feature space and the set M constitutes a sample space used for projection. Then, the projection coefficients are obtained by projecting the sample space to the feature space. Finally, based on the method of PCA, new data are synthesized using the feature space and the projection coefficients extracted from the existing real data. Dynamic changing of the samples in the two sets can synthesize more and more new samples. The number of the samples in the two sets is also decided by our experiments. Therefore, a new synthesized DHV image database containing 8007 subjects is built in our research with 94 original subjects. The experimental results show that the synthesized database reaches a satisfied recognition rate of 97.84%, which indicates the proposed method performs well and would be applicable in the simulation test.

Key words: principal component analysis (PCA), dorsal hand vein (DHV), image synthesis, dynamic space, biological recognition