基于SIFT算法及三角形约束的SAR影像匹配方法

将三角形约束方法引入到合成孔径雷达(SAR)影像匹配中。利用尺度不变特征转换(SIFT)算子生成特征点；采用鲁棒性较好的随机抽样一致(RANSAC)算法剔除错误匹配点，得到更高精确度的同名点；最后利用SIFT算法得到的同名点建立Delaunay三角网。在同名相似三角网内，以三角形重心点作为内插的虚拟同名点，并对虚拟同名点进行归一化互相关(NCC)约束，剔除不满足阈值要求的虚拟同名点对，同时根据内插得到的虚拟同名点建立新的三角网，对三角网进行动态更新，用于获取更多虚拟同名点，直至满足匹配要求。实验结果表明，本文方法能够有效增加匹配特征点数量，提高雷达影像的匹配精确度。

SAR image matching based on SIFT algorithm and triangle constraint

The triangulation constraint method is introduced into Synthetic Aperture Radar(SAR) image matching. This method uses the Scale-Invariant Feature Transform(SIFT) operator to generate feature points. After the SIFT feature points are generated, the RANdom Sample Consensus(RANSAC) algorithm is adopted to eliminate the false matching points, and the higher-precision points with the same name are obtained. Finally, the Delaunay triangulation is established by the same-named point obtained by the SIFT algorithm in the similar triangulation of the same name, the triangle center of gravity is used as the interpolated virtual name of the same name. A Normalized Cross-Correlation(NCC) constraint is performed on the virtual point of the same name, and the virtual point-name pairs that do not meet the threshold requirements are eliminated. Then, a new triangulation network is established according to the virtual same-named point obtained by the interpolation, and the triangulation network is dynamically updated to obtain more virtual names with the same name until the matching requirement is met. The experimental results show that the proposed method can effectively increase the number of matching feature points and improve the matching accuracy of radar images.