Delaunay三角网约束的Harris-SURF图像匹配算法  

Harris-SURF Image Matching Algorithm Constrained by Delaunay Triangulation

作　　者：张贝贝 舒红[1] 江万寿[1] ZHANG Bei-bei;SHU Hong;JIANG Wan-shou(State Key Laboratory of Information Engineering in Surveying,Mapping and Remote Sensing,Wuhan University,Wuhan 430079,China)

机构地区：[1]武汉大学测绘遥感信息工程国家重点实验室

出　　处：《地理与地理信息科学》2019年第6期31-37,I0004,共8页Geography and Geo-Information Science

基　　金：南方电网重点科技项目(GDKJQQ20161187)

摘　　要：Harris算法提取的角点定位精度高,但不具尺度不变性,SURF算法虽具有尺度不变性和旋转不变性,但提取的特征点并非视觉角点。针对此问题,该文提出一种Delaunay三角网约束下的Harris-SURF图像匹配方法。首先,采取阈值评估策略对图像进行SURF粗匹配,利用RANSAC算法进行粗差剔除,得到的匹配点用于构建Delaunay三角网;然后以相似三角形作为约束,将其作为Harris特征点精匹配的限制区域,提高Harris点匹配的可靠性。实验表明,该算法具有匹配准确率高、鲁棒性较好等特点,对无人机影像的匹配效果明显优于其他算法。The Harris algorithm extracts corner points with high precision and without scale-invariant,while SURF algorithm is a method with scale and rotation invariance,but the extracted feature points aren′t visual corner points.Thus,combining the characteristics of the two algorithms,this paper proposed a Harris-SURF image matching method constrained by Delaunay triangle.Firstly,a threshold evaluation strategy is used to extract SURF coarse matching points,then the Delaunay triangulation is constructed by SURF matching points with gross error elimination by the RANSAC algorithm.Secondly,the triangle similarity function is introduced to select matching triangle pairs as constrained area of Harris point fine matching,to improve the reliability of Harris point matching.The experimental results show that the proposed method has better matching accuracy and robustness,it also has higher matching precision in the UAV images,and the matching effect is obviously superior to other algorithms,which has great practical significance of application.

关 键 词：SURF HARRIS 阈值评估策略 DELAUNAY三角网 三角形相似性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]