系统仿真学报 ›› 2017, Vol. 29 ›› Issue (5): 1147-1152.doi: 10.16182/j.issn1004731x.joss.201705029

• 短文 • 上一篇    下一篇

一种基于特征点跟踪的多影像匹配算法

贾开开, 张振杰, 郝向阳   

  1. 解放军信息工程大学,河南 郑州 450001
  • 收稿日期:2015-04-28 修回日期:2015-07-07 出版日期:2017-05-08 发布日期:2020-06-03
  • 作者简介:贾开开(1989-),男,河南商丘,硕士生,研究方向为计算机视觉、机器视觉;张振杰(1988-),男,河南林州,博士生,研究方向视觉导航、计算机视觉;郝向阳(1966-),男,河南济源,博士,教授,博导,研究方向为计算机视觉、摄影测量。

Novel Method on Features Tracking for Multi-image Matching

Jia Kaikai, Zhang Zhenjie, Hao Xiangyang   

  1. Information Engineering University, Zhengzhou 450000, China
  • Received:2015-04-28 Revised:2015-07-07 Online:2017-05-08 Published:2020-06-03

摘要: 在序列影像中,考虑到相邻帧影像的运动随时间变化比较缓慢,可以将光流法用于影像匹配过程中,提出了一种基于特征点跟踪的多影像匹配算法。通过光流法对特征点在帧间进行跟踪,结合匹配判别准则和FB Error算法实现相邻帧影像的快速准确匹配,利用多帧影像间的特征点跟踪实现多影像匹配。实验表明:该算法相比于传统的特征点匹配算法,提高了计算效率并保持了较好的匹配准确率。

关键词: 光流法, 特征点跟踪, 多影像匹配, FB Error

Abstract: Considering the slow motion between adjacent frames in image sequence, optical flow can be applied on image matching. A novel method for multi-image matching based on features tracking was proposed. The algorithm tracked features between adjacent frames, combining with matching strategy and FB Error, to achieve fast and accurate matching. Multi-image matching was finished by tracking features between image sequence. The experimental results show that the novel algorithm improves the efficiency and ensure the accuracy.

Key words: optical flow, features tracking, multi-image matching, FB Error

中图分类号: