相对定向中最优特征点匹配算法的研究与应用  

Best feature point matching method research and application in automatic relative orientation

作　　者：李健[1,2] 刘凤德[2] 赵利平[2] 李靖[2] 张刚[2] 

机构地区：[1]武汉大学遥感信息工程学院,武汉430079 [2]中国测绘科学研究院,北京100039

出　　处：《测绘科学》2010年第4期84-86,共3页Science of Surveying and Mapping

基　　金：国家自然科学基金(40901178);中国测绘科学研究院2008年度基本科研业务费项目(77742);地理空间信息工程国家测绘局重点实验室经费资助项目(200924;200925)

摘　　要：全自动相对定向的关键技术是特征点的自动匹配算法,包括特征点提取算法和特征点匹配算法。本文在分析了几种常用算法的基础上,提出了基于区域特征与角点特征相结合的匹配方法,实验表明该方法精度高、稳定性好,在实际应用中也取得良好效果。Image matching based on feature point is the key technology in automatic relative orientation. Nowadays, feature detectors such as Forstner, Moravec. Harris, SIFF are being applied widely in photogrammetry. What are the respective characteristics of them？ And how about the influence of those method on precision in relative orientation？ The paper introduced the Basic principle of those methods and presented a new matching method based on combining feature detectors and descriptors. The experimental results indicate that it has the advantages of high precision, high reliability. The technique has already been used in practical application and has achieved a satisfactory result, and has a significant reference for research and application in the future.

关 键 词：相对定向 特征提取 最小二乘匹配 可靠性 SIFT 

分 类 号：P23[天文地球—摄影测量与遥感]