基于相对定向的SIFT误匹配点剔除的算法研究
Studying on the algorithm of Relative Oritation in excluding feature errors based on sift
近几年来，基于SIFT算法的图像匹配成为国内研究的热点。SIFT算法在影像特征点提取方面具有独特的优势，但在匹配方面还存在一定得缺陷。匹配的特征点只能代表图像对间的局部关系，有限数量的匹配特征点并不能完全反映图像间的全局关系。在对应的匹配点对之间，存在误匹配或匹配误差较大的点对，由于这些点对的存在将影响匹配的精度。因此，在进行匹配完成之后，探讨误匹配点剔除的方法研究显得就十分的重要。本文提出了一种利用相对定向的方法进行误匹配点剔除的方法策略，并通过vc++上机实验验证其可行性。实验结果表明，此方法具有可行性，取得了预期剔除误匹配点的效果.
In Recent years,Studying on image matching based on SIFT has become a hot topic in china. Though SIFT has advantage in extracting feature, it has limitation in imaging match. The matched points only represent the local relation but not the whole relation. It will have error in matched points and have effect on the precision.So,How to wipe off these error points is very important. In this thesis,the author suggested a method of excluding features errors based on SIFT using Relative Oritation and through VC++ validating its feasibility. The experiment result show that the method got the anticipated purpos. The method got the anticipated purpos.
孙庆华、张宪柱、苏晓蓓
计算技术、计算机技术遥感技术
SIFT影像金字塔相对定向加权迭代
SIFTRelative OritationImaging PyramidWeighted iteration
孙庆华,张宪柱,苏晓蓓.基于相对定向的SIFT误匹配点剔除的算法研究[EB/OL].(2009-09-29)[2025-08-02].http://www.paper.edu.cn/releasepaper/content/200909-826.点此复制
评论