东北大学学报(自然科学版) ›› 2010, Vol. 31 ›› Issue (12): 1689-1692.DOI: -

• 论著 • 上一篇    下一篇

基于尺度、距离、旋转测度的角点匹配算法

贾旭;薛定宇;崔建江;   

  1. 东北大学信息科学与工程学院;
  • 收稿日期:2013-06-20 修回日期:2013-06-20 出版日期:2010-12-15 发布日期:2013-06-20
  • 通讯作者: -
  • 作者简介:-
  • 基金资助:
    国家自然科学基金资助项目(60671050)

Corners matching algorithm based on measures of scale, distance and rotation

Jia, Xu (1); Xue, Ding-Yu (1); Cui, Jian-Jiang (1)   

  1. (1) School of Information Science and Engineering, Northeastern University, Shenyang 110004, China
  • Received:2013-06-20 Revised:2013-06-20 Online:2010-12-15 Published:2013-06-20
  • Contact: Jia, X.
  • About author:-
  • Supported by:
    -

摘要: 针对基于模板灰度相似性测度的匹配方法抗旋转性差的缺陷,依据真实匹配角点与邻域内其他角点位置关系存在拓扑不变性,提出了一种基于尺度、距离、旋转测度的角点匹配方法.该方法首先利用相位相关法对角点进行初始匹配,而后对每对候选匹配角点进行基于尺度、距离、旋转测度的计算,利用计算后的测度函数值来判断初始匹配是否为正确匹配.实验验证了提出的角点匹配方法的匹配结果要明显好于直接基于模板灰度相似性测度的匹配方法.

关键词: 双目视觉, 角点匹配, 相位相关, 测度计算, 拓扑不变性

Abstract: Because there is a topological invariance of location between true corner to be matched with other corners in neighborhood, a new corner matching algorithm based on measures of scale, distance and rotation is proposed to solve the drawbacks of conventional matching algorithms based on the similarity measure of gray levels. In this algorithm the corners are matched initially according to the phase-correlation, and every couple of corners to be matched is computed according to the measures of scale, distance and rotation. Then, whether a match is right or wrong is judged by the measure function values given by the computation. Some experimental results verified that the matching algorithm proposed is better than other conventional algorithms for matching corners.

中图分类号: