Computer Engineering and Applications ›› 2012, Vol. 48 ›› Issue (14): 168-170.

Previous Articles     Next Articles

New image matching algorithm based on local invariant features

WU Wenhuan1, LI Qian1, JIANG Zetao2, YANG Jun1   

  1. 1.Department of Computer Science, Zhoukou Normal University, Zhoukou, Henan 466001, China
    2.School of Information Engineering, Nanchang Hangkong University, Nanchang 330063, China
  • Online:2012-05-11 Published:2012-05-14

基于局部不变特征的图像匹配算法

吴文欢1,李  骞1,江泽涛2,杨  俊1   

  1. 1.周口师范学院 计算机科学系,河南 周口 466001
    2.南昌航空大学 信息工程学院,南昌 330063

Abstract: Aiming at the image matching in the field of computer vision, this paper presents a new matching algorithm based on local invariant features. Feature points are detected by difference of Gaussian. The the Haar-wavelet responses within a feature point neighbourhood are projected into four directions, and then a 64-dimensional vector is generated for describing the feature point. Matching pairs are determined by using the nearest neighbor distance ratio. Experimental results show that the proposed algorithm is not only rapid and robust, but its matching rate is higher than PCA-SIFT, SURF and MSOP.

Key words: image matching, difference of Gaussian, feature vector, Haar-wavelet

摘要: 针对计算机视觉领域中的图像匹配问题,提出一种新的基于局部不变特征的匹配算法。使用高斯差分检测特征点,将特征点领域内Haar小波响应投影到四个方向轴上,进而生成一个用来描述特征点的64维向量,采用最近邻距离比进行特征匹配。实验结果表明,该算法不仅快速、稳定,而且匹配准确率比PCA-SIFT 、SURF、MSOP高。

关键词: 图像匹配, 高斯差分, 特征向量, Haar小波