顾及拓扑关系的立体影像直线特征可靠匹配算法

发布时间：2018-01-19 23:34

本文关键词： 直线匹配直线组匹配拓扑关系核线约束单应矩阵　出处：《测绘学报》2017年11期 　论文类型：期刊论文

【摘要】：针对单直线匹配过程中缺乏考虑邻近直线特征之间关系,纹理断裂处单一直线描述符的弱可靠性,提出了一种顾及拓扑关系的立体影像直线特征可靠匹配算法。该算法首先根据直线间距离、角度等基本拓扑关系对参考影像、搜索影像上提取的直线进行编组;然后将编组得到的直线组作为匹配基元,充分利用直线特征组内的拓扑关系,依次采用核线约束、单应矩阵约束、象限约束、不规则三角形区域灰度相关约束对其进行匹配;最后将同名直线组分裂为两对同名单直线、并对分裂后的结果进行整合、拟合、检核等后处理,得到"一对一"的同名直线。选取典型纹理特征的航空影像和近景影像进行参数分析及直线匹配试验,结果表明,本文算法能获取可靠的直线匹配结果。
[Abstract]:In the process of single line matching, the relationship between adjacent line features is not considered, and the weak reliability of single line descriptor at the break of texture is discussed. In this paper, a reliable matching algorithm of stereo image line features considering topological relation is proposed. Firstly, the reference image is grouped according to the distance and angle of the lines. Then, the straight line group is used as the matching primitive, and the topological relation in the line feature group is fully utilized. The kernel line constraint, monoclinic matrix constraint and quadrant constraint are adopted in turn. The irregular triangle region is matched by gray correlation constraints. Finally, the line group of the same name is divided into two pairs of the same list line, and the results after the split are integrated, fitted, checked and so on post-processing. A one-to-one straight line with the same name is obtained. The parameter analysis and line matching test of the typical texture feature aerial image and close-range image are carried out. The results show that the proposed algorithm can obtain reliable straight line matching results.
【作者单位】：西南交通大学地球科学与环境工程学院;辽宁工程技术大学测绘与地理科学学院;深圳大学智慧城市研究院;
【基金】：国家自然科学基金(41631174;41101452) 辽宁省教育厅科学研究一般项目(LJYL010)~~
【分类号】：TP391.41
【正文快照】： 直线特征是人类视觉信息中最显著、最直观、最具有代表性的特征。城区建筑物轮廓边缘、屋顶和立面都蕴含着大量的直线特征,包含丰富的语言信息和几何意义,能够较好地表达建筑物结构特征,因此,将直线特征作为匹配基元并实现其可靠匹配是基于影像精细三维建模的基础[1-4]。目前,

【相似文献】