一种改进的喀斯特山区无人机影像匹配算法-《计算机技术与发展》

文章信息/Info

Author(s):: DAI You-lin¹; WANG Xiao-hong^{2 *} ; LI Wan-hua¹; 1. School of Mining,Guizhou University,Guiyang 550025,China;
2. School of Forestry,Guizhou University,Guiyang 550025,China

摘要:: 喀斯特山区复杂的地形条件使得该地区的无人机影像匹配较困难。针对如何提高该地区影像匹配的正确率和时效性,提出一种组合特征的改进算法。该算法采用 AKAZE 算法对该地区的无人机影像进行特征检测,并运用 DAISY 描述子对提取的特征进行描述,最后采用随机抽样一致性算法对匹配进行错误剔除。对该算法与 AKAZE 算法和 SURF 算法进行了对比试验。试验结果表明,在喀斯特山区的无人机影像匹配中,与 AKAZE 算法相比,这种组合特征的改进算法在保持同等总匹配数的情况下,正确率提高 14% 左右,匹配总耗时减少 15% 以上,单个正确匹配耗时减少 30% 以上,较大提升了匹配的效率。该算法相较于 SURF 算法,有更多的总匹配数,正确率提高 14% 左右,单个正确匹配耗时减少 10% 左右。这种组合特征的算法相较于 SURF 算法和 AKAZE 算法更适合喀斯特山区的无人机影像匹配。

Abstract:: The complex terrain conditions in Karst mountains make it difficult to match UAV images in this area. Aiming at how to improve the accuracy? ? ?and timeliness of image matching in this area,an improved algorithm for combining features is proposed. Firstly,the AKAZE algorithm is used to perform feature detection on UAV images in the area, and then the DAISY descriptors are used to describe the characteristics extracted. Finally the RANSAC algorithm is used to eliminate the matching errors. The algorithm is compared with AKAZE algorithm and SURF algorithm. The test shows that in the UAV image matching in the Karst mountain area,compared with the AKAZE algorithm,the proposed algorithm can increase the accuracy by about 14% while maintaining the same total number of matches,and reduce the total matching time by more than 15% and a single correct matching time by more than 30% ,which greatly improves the efficiency of matching. Compared with the SURF algorithm,the proposed algorithm has more total matches,the accuracy rate is increased by about 14% ,and the time for a single correct match is reduced by about 10% . Compared with SURF algorithm and AKAZE algorithm,this combined feature algorithm is more suitable for UAV image matching in Karst mountain area.