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.
出处
《测绘科学》
CSCD
北大核心
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)
关键词
近景影像
SIFT算子
极线约束
单应矩阵
RANSAC算法
close-range image
Scale Invariant Feature Transform （or SIFT） operator
epipolar constraint
homograph matrix constraint
RANSAC （random sample consensus） algorithm
