摘要
提出一种基于改进的DCCD(double-cirele-based corner detector,)和SIFT(scale invariant feature transform)描述符的影像匹配方法。在特征点检测阶段,首先采用改进的DCCD快速检测影像上的关键点,然后确定关键点的主方向,生成特征点。在特征点描述阶段,采用SIFT描述符描述特征点。在特征点匹配阶段,分别采用BBF(best bin first)算法和RANSAC(随机采样一致性)算法进行特征点粗匹配和误匹配特征点剔除。实验结果表明,与基于Harris角点和SIFT描述符的影像匹配方法相比,该方法在匹配速度和准确率方面得到了提高。
An image matching method based on improved DCCD and SIFT descriptor is proposed.With the proposed method,key points are detected rapidly using improved DCCD,and then the orientation of these key points is determined to form the feature points from these key points.And SIFT descriptor is used to describe these feature points.As for feature point matching,BBF algorithm and RANSAC algorithm are used for rough matching andeliminating false matching respectively.The experimental results show that the proposed method possesses faster matching speed and higher matching accuracy compared to the image matching method based on Harris corner and SIFT descriptor.
出处
《武汉大学学报（信息科学版）》
EI
CSCD
北大核心
2015年第12期1613-1617,1645,共6页
Geomatics and Information Science of Wuhan University
基金
国家自然科学基金资助项目(41371365)~~
