东北大学学报（自然科学版）

东北大学学报(自然科学版) ›› 2008, Vol. 29 ›› Issue (11): 1621-1624.DOI: -

• 论著 • 上一篇    下一篇

一种基于PCA的工件图像匹配方法的研究

张金萍;刘杰;李允公;倪洪启;   

  1. 东北大学机械工程与自动化学院;沈阳化工学院机械工程学院;
  • 收稿日期:2013-06-22 修回日期:2013-06-22 出版日期:2008-11-15 发布日期:2013-06-22
  • 通讯作者: Zhang, J.-P.
  • 作者简介:-
  • 基金资助:
    国家自然科学基金资助项目(50775029)

PCA-based method for image matching of workpiece

Zhang, Jin-Ping (1); Liu, Jie (1); Li, Yun-Gong (1); Ni, Hong-Qi (2)   

  1. (1) School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110004, China; (2) School of Mechanical Engineering, Shenyang Institute of Chemical Technology, Shenyang 110142, China
  • Received:2013-06-22 Revised:2013-06-22 Online:2008-11-15 Published:2013-06-22
  • Contact: Zhang, J.-P.
  • About author:-
  • Supported by:
    -

RichHTML

1

PDF (PC)

543

摘要： 为提高工件图像匹配的计算效率提出了一种基于主元分析(PCA)的新方法.该方法将模板图像视为一高维空间点集,对其进行PCA处理,在确定能够保证信息保留率的主元个数后,得到各元方向上投影值的最大最小值及与其对应的模板图像中的列的序号.在匹配时,只需抽取搜索区域中相同序号的列向相应的主元方向作投影运算,从而构成投影值向量,进而计算该向量与预处理中所得的最大最小值构成向量间的距离,从而根据距离的大小判断是否匹配.匹配中的寻优工具使用一种改进的遗传算法.对工件匹配的实验验证说明,所提方法具有较高的寻优速度和精度,且在待匹配图像中混有较高噪声的情况下也能得到较好的结果.

关键词: 工件识别, 图像匹配, 主元分析, 主元方向, 遗传算法

Abstract: To speed up the computation for workpiece image matching, a novel PCA (principal component analysis) method is proposed taking the template image as a multi-dimension set of spatial points, to which the PCA is made. Then, the number of principal components is determined to ensure the information reserving rate, thus obtaining the maximum and minimum projection values (MM-vectors) in all directions of those components as well as the serial numbers of rows in corresponding template images. While matching, only the principal component directions corresponding to the rows of the same serial numbers in searching area are needed to be extracted for the computation of projection value so as to form the vectors of projection values. The distances between MM-vectors and the vectors formed by projection values are computed, and whether the two kinds of vectors are matchable should be in accordance to those distances. During the matching an improved genetic algorithm is used for optimization. The matching results of workpiece images show the high optimizing speed and precision of the proposed method, even if the images to be matched are mixed with high decibel noise.

中图分类号: