【摘要】 针对喀斯特地貌地形起伏大、沟壑纵深、影像局部存在阴影,该地区无人机遥感影像匹配难度大,该文提出一种基于AKAZE特征和RANSAC算法的喀斯特地貌无人机遥感影像的快速匹配方法。该方法采用AKAZE特征检测算法提取喀斯特地貌无人机遥感影像的特征点,能够较为精简地描述出影像的特征,影像匹配时间缩短,匹配速度提高明显;由于预匹配存在错误点对,利用单应性矩阵的RANSAC算法进行精准匹配,剔除错误匹配点对,同时得到最优的单应性矩阵。实验结果表明,该方法的匹配耗时和匹配正确率较优于经典的SURF和BRISK算法,该匹配方法更适合于喀斯特地貌无人机遥感影像匹配。更多还原

【Abstract】 In Karst Landform,in view of the large terrain,deep ravines and partial shadow,the UAV remote sensing image matching is difficult.In this paper,a fast matching method of UAV remote sensing images in Karst Landform based on AKAZE features and RANSAC algorithm is proposed.This method uses AKAZE feature detection algorithm of feature point extraction in Karst Landform of UAV remote sensing images,which can be more streamlined to describe feature of image,image matching time is shortened,the matching speed is obviously improved.There is an error point pair due to the prematching,so this paper will use RANSAC algorithm of homography matrix for accurate matching,eliminating the false matching points at the same time on, get the optimal homography matrix.Experimental results show that the matching time and matching rate of this method are better than those of classical SURF and BRISK algorithm.This matching method is more suitable for UAV remote sensing image matching in Karst Landform.更多还原