一种针对近景影像的匹配算法  

A matching algorithm for close-range image

作　　者：卜范秋[1] 宋伟东[1] 王伟玺[1,2] 

机构地区：[1]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000 [2]深圳市规划国土发展研究中心,广东深圳518034

出　　处：《测绘科学》2013年第6期143-146,194,共5页Science of Surveying and Mapping

基　　金：国家自然科学基金项目"面向复杂建筑物部件的地面激光扫描点云与近景影像混合三维建模方法研究"(40901222);"面向多角度数字航空影像的多基元多层次可靠匹配方法"(41101452)

摘　　要：近景影像的匹配要比航空摄影影像的匹配存在更大的难度。本文提出一种针对近景影像特征提取与匹配的研究方案:利用SIFT算子提取影像的特征点并进行初始匹配,获得的精度较高的初始匹配点对计算约束条件,然后利用约束条件重新引导匹配,实现约束匹配的过程;并利用RANSAC算法消除误匹配,达到精匹配的效果。实验表明:本文提出的匹配方法适用于近景影像,并取得了很好的效果。Close-range image matching is more difficult than airborne/spaceborne image matching. This paper proposed an algorithm for feature extraction and matching for close-range image. First, it made initial matching by extracting feature points based on SIFT operator, then calculated the constraint condition （homography matrix, fundamental matrix） by high-precise matching points ob- tained in initial matching, and constrained the matching by homography matrix and fundamental matrix constraint, finialiy eliminated false matches by RANSAC algorithm and obtained high-precise matching results. Experimental results showed that the proposed algo- rithm could be suitable for close-range image with reliable matching results.

关 键 词：近景影像 SIFT算子 极线约束 单应矩阵 RANSAC算法 

分 类 号：P234.1[天文地球—摄影测量与遥感]