• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2014, Vol. 36 ›› Issue (09): 1765-1769.

• 论文 • 上一篇    下一篇

基于K-means聚类和RANSAC的图像配准算法研究

王天召,徐克虎,陈金玉   

  1. (装甲兵工程学院控制工程系,北京 100072)
  • 收稿日期:2012-11-26 修回日期:2013-03-25 出版日期:2014-09-25 发布日期:2014-09-25
  • 基金资助:

    总装院校创新基金(2012CJ42)

Research on the image registration algorithm
based on K-means clustering and RANSAC      

WANG Tianzhao,XU Kehu,CHEN Jinyu   

  1. (Department of Control Engineering,Academy of Armored Force Engineering,Beijing 100072,China)
  • Received:2012-11-26 Revised:2013-03-25 Online:2014-09-25 Published:2014-09-25

摘要:

针对图像配准中特征点匹配方法存在实时性不高和精度低的问题,提出了一种基于Kmeans聚类和RANSAC的图像配准算法。该算法根据匹配点对距离和方向特征的视差约束条件,首先利用Kmeans聚类对匹配点对进行预处理,剔除大部分错误匹配点,然后利用RANSAC进行二次优化,实现了图像的快速和精确配准。实验结果表明,该算法不仅提高了图像配准的精确度,而且提高了图像配准的速度。

关键词: 图像配准, 特征点匹配, K均值聚类, 随机样本一致

Abstract:

Aiming at the problems of less realtime and low precision of feature matching in image registration, an image registration algorithm based on Kmeans and RANSAC is proposed.Based on the parallax constraint of distance and angle features of matching point pairs,the algorithm firstly uses Kmeans clustering to proprocess matching point pairs in order to filter false matching point pairs, and secondly adopts RANSAC to optimize the matching point pairs so as to realize fast and precise image registration.The experimental results indicate that the algorithm improves both the speed and the precision of image registration.

Key words: image registration;feature point matching;K-means clustering;RANSAC