Research On The Method Of Geographic Polygonal Feature Matching Based On Objective Clustering

One of the most important issues in GIS is sharing the spatial data with others, in which the key problem is to identify the counterpart map elements from different data sources and create the logical relationship between them. However, there are two main problems in current studies, one is that the counterpart elements matching models are constructed artificially, so it can’t guarantee the matching model is an optimal model. The other is that the previous work didn’t consider finding the candidate matching set for multiple elements simultaneously. In order to solve those problems, we study the polygonal elements and doing some experiments.For the first problem, genetic algorithms is used to achieve optimal selection of matching measures and build a matching model. In other words, select the measures to construct the matching model in a random way. And then, a highly fitted matching model after several times of iterated matching using the selection, cross and mutation genetic operations. For the second problem, the objective clustering of polygonal elements is presented for the rebuilding the mapping relationships of overlap the cluster regions and the selection among multiple candidate matching features.The data of residents, administrative area and drainage network, as three typical subjects are selected. The main conclusions are followed:(1) Our method which based on the genetic algorithms could select appropriate matching measures to build the matching model on the basis of specific geographic data. The matching results show that the method proposed by this paper identifies most of the counterpart polygonal elements effectively.(2) The method based on the objective clustering can select out the candidate matching set for multiple elements simultaneously. In the same time, we increase the matching efficiency by narrowing down searching the candidate matching scope for single element.Specifically, the main work and results of this paper can be summarized as follows:1. In order to take the place of manual process of building matching models, we use the genetic algorithms.2. Selecting the candidate matching set for multiple elements simultaneously.The meaning of this study is combining the genetic algorithm and the matching of counterpart elements, so we could build the matching model for counterpart elements automatically and lay the foundation for other researchers. On the other hands, identifying the candidate matching set by the objective clustering, also provides a new solution for this problem.