计算机与数字工程2019，Vol.47Issue(5)：1233-1239,7.DOI:10.3969/j.issn.1672-9722.2019.05.043
无人机自主定位中的仿射二进制特征匹配
Binary Feature Matching in UAV Autonomous Location
摘要
Abstract
Because there is big positioning error when using GPS in the indoor or in close buildings,using visual to autono?mously locate can solve this problem with a cheaper and more accuracy way. Different from most of the traditional affine matching al?gorithm which use floating-point feature descriptors,the binary character descriptors have great advantages on memory usage and time consumption. Therefore,this paper proposes a fast-affine algorithm based on virtual perspective and binary features image matching algorithm. First,the binary features of the images to be matched are extracted,and homograph matrix is generated through rough matching. Then,the homograph matrix is evaluated to determine whether generate the intermediate image with homograph ma?trix or generate simulation image with virtual perspective. Finally,the results are gotten. Simulation image with virtual perspective means generating images according to simulating the position of camera optical axis. Experiments show that the fast-affine algorithm proposed in this paper can well meet the requirements of different Angle matching between ground and air images,and its comput?ing speed is 30 times faster than ASIFT algorithm and 5 times faster than AORB algorithm.关键词
无人机/城市环境/地空图像匹配/仿射变换/ORB特征/虚拟视角Key words
UAV/urban environments/air-ground image matching/affine/ORB/virtual perspective分类
航空航天引用本文复制引用
张含茜,石朝侠..无人机自主定位中的仿射二进制特征匹配[J].计算机与数字工程,2019,47(5):1233-1239,7.基金项目
国家自然科学基金面上项目"基于场景-拓扑的多机器人协作环境搜索研究"(编号:61371040)资助. （编号:61371040）
