【摘要】 针对高原山区地形条件下如何快速有效地进行无人机影像匹配的问题,提出一种适应高原山区地形条件的综合算法,其集成了方向性的高速特征检测(FAST)与旋转不变特征描述(ORB)算法、二进制鲁棒不变的可扩展关键点(BRISK)描述子、基于网格的运动统计算法和随机抽样一致性算法(RANSAC)模型约束改进的方法 .使用ORB算法中的FAST通过对像素灰度值的快速比较来检测特征点,使用BRISK描述子生成适合影像特征的描述子;采用集成暴力匹配算法和基于网格的运动统计算法,对特征点进行第一次约束优化提取;运用RANSAC对匹配点对进行第二次约束优化提取,以剔除不正确的匹配点对,提高影像匹配的准确性.结果显示,在以恐龙谷为例的高原山区复杂地形影像匹配中,改进算法在计算时间方面具有优势,且能够提取许多精确的特征点对.更多还原

【Abstract】 Aiming at the problem of how to match unmanned aerial vehicle images quickly and effectively under the terrain conditions of mountainous plateau area, we presented a comprehensive algorithm adapted to the terrain conditions in such areas. The algorithm integrates oriented features from accelerated segment test(FAST) and rotated binary robust independent elementary features(ORB) algorithm and binary robust invariant scalable keypoints(BRISK) descriptor, grid-based motion statistics, and random sampling consistency(RANSAC) model constraint improvement methods. The detailed process is as follows: we used FAST feature extraction in the ORB algorithm to detect feature points by quickly comparing pixel gray values, and used BRISK descriptor to generate a descriptor suitable for image features. The brute-force matching algorithm and grid-based motion statistics algorithm were used to extract the feature points for the first time. The RANSAC algorithm was used to extract the matching point pair for the second time to eliminate the incorrect matching point pair and improve the accuracy of image matching. The experimental results showed that the improved algorithm has significant advantages in computing time, and can extract many accurate feature point pairs in the complex terrain image matching of the plateau mountain area, with the Dinosaur Valley as an example.更多还原