基于Hough变换的扫描地形图几何纠正算法研究  

A Geometric Rectification Algorithm of Scanned Topographic Map Based on Hough Transform

作　　者：李关义[1] 林友明[1] 

机构地区：[1]中国科学院中国遥感卫星地面站

出　　处：《遥感技术与应用》2006年第2期109-114,共6页Remote Sensing Technology and Application

摘　　要：扫描的地形图存在一定的误差,可采用方里网精纠正方法对其进行几何纠正。这需要大量的控制点,一般采用图中方里网格线交点作为控制点。针对地形图网格点的提取,对Hough变换进行了改进,即采取预测网格点的方法确定搜索区域,用Hough变换提取方里网格点,然后把所有检测出的方里网格点组成三角网格,用仿射变换对地形图进行分三角网格精纠正。实验表明,该方法检测精度高,只需要少量的人工干预,即可实现所有网格点的子像元级提取。This paper gives a thorough study on geometric rectification of scanned topographic map. There are many distortions caused by some reasons in scanned topographic maps. So they must be rectified before we use them. To reduce distortions effectively, the method of triangulation with grid points could be used usually for geometric rectification and lots of control points are required. But it＇s too hard to extract grid points manually. In order to identify all grid points automatically, a novel geometric rectification algorithm of scanned topographic map is presented in this paper, which is based on the improved hough transform. At first, we calculate the initial positions of grid points approximately using information of topogtaphy map, and make a ＂十＂ shape search area around each grid predicted point. And then, hough transform is used to detect grid lines in every search area and the intersections of grid lines are calculated. Finally, the map is rectified triangularly by affine transform, with intersections detected by hough transform as reference points. Experimental results have showed that this method had the high precision performance to realize the sub-pixel precision extraction of grid points and no need to find reference points one by one manually. And it＇s also faster than traditional manual method.

关 键 词：HOUGH变换 直线检测 子像元精度 扫描地形图纠正 控制点提取 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]