文中提出一种基于邻域投票和改进的 Harris-SIFT特征的低空遥感影像匹配方法.首先用NMS算法提取多尺度的 Harris-SIFT特征并对其进行方向描述 ,然后根据最近邻与次近邻特征点距离之比确定初始匹配点对 ,最后通过近邻域选择投票的方法剔除候选点中的虚假匹配点 ,进而实现低空遥感影像的配准.实验表明该算法在获得充足匹配点且保证匹配精度的同时 ,明显提高影像匹配的效率.%Based on the neighborhood selection algorithm and improved Harris-SIFT features ,a method for low-altitude remote sensing matching is presented .Firstly ,it extracts Harris-SIFT features using NMS algorithm and generates feature descriptors ,then uses ratio method to get initial matching .Finally ,it uses neighborhood selection algorithm to eliminate errors and achieves accurate matching .Experiments show the algorithm can get adequate matching points ,ensure accuracy and improve the matching efficiency .
展开▼