基于动态规划和Hough变换的遥感影像与GIS矢量数据匹配方法

摘要

Due to the essential difference between remote sensing images and GIS vector data, and some changes at position and scale after remote sensing images and GIS vector data approximate overlay, it was difficult to establish a correspondence between them. In the paper, a from-coarse-to-fine matching method with multiply stages was proposed. This method firstly taking window automatic feature extraction as the base, the optimal partial matching points and the line segments to be matched as premise, performed the polylines whole matching by dynamic programming; then the transform parameters of geometric model were calculated with corresponding points which were gained in the preceding step, to achieve rough matching; ultimately, making use of Hough transform, went on feature extraction and feature matching again in synchrony. This process regarding the rough matching results as the initial state, effectively improved the accuracy of matching. Experimental results indicated this method was effective and robust.%由于遥感影像与GIS矢量数据性质不同,并且两者叠加后存在一定的平移、旋转和尺度变化等原因,准确建立两者之间的对应关系是一个难点.为此,提出一种多阶段、由粗到精的匹配方法.该方法在遥感影像特征自动提取的基础上,以局部匹配优选的待匹配结点和待匹配线段为前提,利用动态规划匹配方法,进行折线整体匹配；然后利用匹配的同名点,解算几何模型变换参数,实现二者粗配准；最后以粗配准结果作为初值,利用Hough变换再次同步进行折线特征提取和特征匹配,有效提高匹配的精度.实验证明该方法的有效性和稳健性.