测绘科学技术学报2016，Vol.33Issue(4)：431-435,5.DOI:10.3969/j.issn.1673-6338.2016.04.020
一种P OS数据辅助多视角倾斜航空影像匹配方法
A POS Supported Matching Method for Multi-View Oblique Aerial Images
摘要
Abstract
A POS supported matching method for multi-view aerial oblique images is put forward. First of all, the POS data is used to rectify the oblique images by epipoliar geometry constraint. Then SIFTGPU algorithm is used for feature matching on the rectified images. According to the matching results, the parallax of horizontal and verti-cal direction is computed to get their approximate overlapping area. Thirdly, the block feature matching is appiled to the overlap areas and the constraint conditions such as nearest neighbor distance ratio, parallax constraints, RANSAC algorithm are applied to eliminate false matches. In the course of time, match results are counted through POS data back to the original images. The result of experiment shows that matching points can be quickly obtained more in quantity and more uniform in distribution compared to the conventional image matching method when the POS data is applied to the multi-view aerial oblique image matching.关键词
倾斜摄影测量/影像匹配/外方位元素/特征匹配/近似核线影像Key words
oblique photogrammetric/image matching/exterior orientation elements/feature matching/approxi-mate epipolar image分类
天文与地球科学引用本文复制引用
赵明衍,戴晨光,狄亚南,仇多兵,张伟政..一种P OS数据辅助多视角倾斜航空影像匹配方法[J].测绘科学技术学报,2016,33(4):431-435,5.基金项目
国家海洋公益基金项目(201305002) （201305002）
国家自然科学基金项目(41571336) （41571336）
中央高校基本科研业务费专项(3132015002 （）
3132014034 （）
3132016003) （）
中央高校基本科研业务费创新团队(3132014302)。 （3132014302）
