低空无人机影像匹配流程设计的优化  

Optimization of Low Altitude UAV Image Matching Design Process

作　　者：姚艳 YAO Yan(Shanxi Basic Geographic Information Institute, Taiyuan Shanxi 030001, China)

机构地区：[1]山西省基础地理信息院,山西太原030001

出　　处：《北京测绘》2019年第4期423-426,共4页Beijing Surveying and Mapping

摘　　要：针对低空无人机倾斜摄影影像匹配精度不高的问题,提出了以SIFT-BRISK算子为基础的影像匹配工作流程。首先,为相邻影像同步建立多尺度影像空间,在不同尺度上检测极值点;然后,根据三维二次函数剔除边缘极值点,确定为影像的匹配点;最后,按照高斯滤波理论,对每个匹配点进行描述,相邻影像各自生成各自的匹配点集合,当两影像匹配点描述相同时,则实现特征点匹配,进而实现相邻影像的高精度匹配。在三维倾斜摄影场景项目中,此方法有效地降低了匹配错误率,提升了匹配效果。Aiming at the problem of low matching accuracy of tilted photographic images of low altitude UAV, a workflow of image matching based on SIFT-BRISK operator is proposed. Firstly, multi-scale image space is established for adjacent images synchronously, and extremum points are detected on different scales. Then, according to three-dimensional quadratic function, the edge extremum points are eliminated and the matching points are determined. Finally, according to the theory of Gauss filtering, each matching point is described, and each adjacent image generates its own set of matching points. When the descriptions of two matching points are the same, the feature points are matched, and then the high-precision matching of adjacent images is realized. This method can effectively reduce the matching error rate and improve the matching effect in the three-dimensional tilt photography scene project.

关 键 词：SIFT-BRISK算子 影像匹配 流程设计 

分 类 号：P231[天文地球—摄影测量与遥感]