利用几何矩和叠置分析进行多尺度面要素自动匹配

针对地图中重要的面要素的自动匹配问题进行了研究。由于在实际情况中,多尺度同名面要素在位置上可能偏差较大,又存在非一对一的匹配关系,现有的方法均难以进行有效匹配。为解决这个问题,首先提取面要素的全局形状特征,并给出了一种快速获取候选匹配面要素组合的方法,然后利用几何矩获取面要素组合的质心,在将待匹配面要素与组合面要素的质心相互重合后,通过计算重叠相似度来判定多尺度面要素之间的匹配关系。实验表明,本文方法即使在同名面要素存在较大位置偏差,甚至非均匀位置偏差时,仍可有效地处理面要素的一对一和一对多两种匹配关系,从而拓展了面要素匹配方法的适用范围,降低了对外界条件的要求和限制。

An Algorithm for Muti-Scale One-to-many Areal Feature Matching Based on Geometry Moments and Overly Analysis

This paper carries out research on technique of matching areal features which is very important in vector maps.In many actual situations,muti-scale homonymous areal features may differ greatly in their locations and there is 1∶1 or 1∶N corresponding relationship between them.In order to solve such a problem,an algorithm based on geometry moments and overlay analysis is proposed.This algorithm firstly acquire candidate matching combined areal features based on presupposition that homonymous areal features should have similar global shape feature,such as same area or similar bound rectangle,and then centroid of areal features group is obtained through computing geometry moments.After making centroid of matching areal features concide with each other,the matching relationship of areal features can be determined through computing overlapping similarity.The experiments indicate that the proposed algorithm can still get a satisfactory result even when homonymous features differ greatly in their locations and their matching relationship is one-to-one or one-to-many.