基于Harris和SURF的遥感图像匹配方法研究

Remote Sensing Image Matching Method Based on Harris and SURF

针对多源遥感图像匹配正确率低的问题,本文首先采用点空间约束的Harris角点检测算法,得到分布比较均匀的角点;接着构建Voronoi图进行图像分块;然后应用分块SURF特征点检测和匹配得到仿射变换参数;再利用灰度积相关算法实现同名点搜索;最后辅以两点对空间约束剔除误匹配。实验结果表明,本文采用的基于Harris和SURF的方法在遥感图像匹配正确率和效率上优于SURF算法。

For the low correct rate of multi-source remote sensing image matching,the Harris corner detection algorithm iswas firstly used in the paper to get uniformly distributed corner,and;then the Voronoi diagram iswas built to obtain image blocks.;after that,Iin order to get affine transformation parameters,blocking of SURF feature-point detection and matching hasd to be done.;In addition,the same-site point search would be realized with gray-scale plot relevant algorithm.;Finally,two point-to-point space constraints arewas made to remove mismatching.The experimental results showed that Harris and SURF method in the paper arewas better than the SURF algorithm in remote sensing image matching accuracy and efficiency.