The Research On Areal Feature Matching Among The Conflation Of Urban Geographic Databases

The capture and maintenance of spatial data takes a lot of time and hard sledding, however, usually different departments take in charge of these spatial data covering the same areas in reality, which result in lots of duplicate data. As the apparatus, observation approaches and department's interest are different. These duplicate data differ in positional accuracy, and often data content as well as attribute information are also different. In order to making full use of existing information, cutting down high spending on data collection, GIS departments often need to incorporate map data which from various sources, and then Map Conflation appears. And feature matching is the key component of Map Conflation.At present, in large-scale urban geographic databases, areal feature often possess big proportion, construction or tenement in such as house property map, cadastral map, topographic map, they are map-user's interest as well as the expression of map's mainly contents. Therefore, this paper focus on the areal feature matching in large-scale urban geographic databases, the main contents and conclusion of this paper is listed as followed:1) The concept of geographic databases Conflation has firstly been discussed as well as its study category, workflow, typical applications. The author has also researched relations and differences between it and related existing technology, has been done some analysis and summarization. This paper has pointed out that an idealized map conflation should component-based solution, Based the requirements of user, the model can be extended. And considered that today's map conflation should include data preprocessing, map alignment, feature matching and checking, user interaction, data transferring.2) The content of map alignment has been researched, which is to serve subsequent feature matching. It is mainly focus on analyzing the development of Rubber-Sheeting, as well as its principle and arithmetic application, and make it test. The test has indicated that should select the identical node as reference point, and the reference point should be local in the dense of features. The more reference points, the result of spatial discrepancy elimination is better.3) The technology of feature matching has been studied as well as its various methods and characteristics. The feature matching could be classified into three types according the similarity measures of spatial feature. They are geometric method, topological method, attribute method. Furthermore, this paper discusses the similarity of areal feature, and characteristics of applications in these matching approaches.4) This paper has been carried out an experiment study about areal feature matching in large-scale urban geographic databases, and discussed the technology of ArcEngine, it present a geometric matching arithmetic use polygon centroid combine some matching criteria. Use pointin polygon rule to find polygon correspondence, and carry out matching checking unite the area and density of polygon, then it can be judge the information of feature matching. It realized an arithmetic about raster polygon centroid, and this paper proposes use ArcGIS9.0's polygon centroid, it both carry out areal feature matching experiment use these polygon centroid each other, and it analyzed the lastly feature matching result.The experiment has indicated that the geometric matching arithmetic use polygon centroid combine some matching criteria is an effective method. This method not need rely simply on the vertex or the line along boundary carry on feature matching, and predigest the process of feature matching, the method can solution the non-one-to-one matching. As can use diversified matching criteria, so the method has expansibility, easy manipulation and is flexible. Otherwise, if most areal feature is regular in the map, then it can use ArcGIS9.0's polygon centroid, the speed of feature matching is rapid, and facility of use. The result of matching is almost the same of raster centroid;if most areal feature is irregular in the map, such as lake, then use raster centroid is more effective, and is more appropriate.