Residential Area Matching Based On Probability Relaxation Method

In order to maintain the reality of geospatial information data,updating the existing map data is an important strategy.Map update includes map matching,change recognition,update data and other processes.Among them,map matching is the basis and important content of map update.Map matching mainly uses information such as geometry,topology,semantics and other information to measure the similarity between the two residential areas to find matching pairs that are targets with the same name,and use the obtained matching relationship to compare and detect change information.Traditional matching methods mainly take advantage of the similarity of their own spatial characteristics,and choose a residential area with the highest similarity to establish a matching relationship.These methods cannot select the best matching object for each object from a global perspective,and cannot match.Comprehensive identification of the relationship.And due to different data sources,mapping time and other factors,there may be multiple correspondences between residential areas.Therefore,it is necessary to start from the overall thinking,comprehensively consider the spatial relationship similarity of residential areas and the spatial similarity of neighboring objects,and combine the global optimal theory to solve the matching problem of residential areas and provide guarantee for map data update.The specific research content is as follows:1)Theoretical basic research.The basic theories and methods in the matching of residential areas are studied,which lays a foundation for the establishment of subsequent matching models.Including spatial similarity theory,global optimization theory,probability relaxation method,and the spatial distribution pattern of residential areas and the corresponding relationship in matching.2)The establishment of a matching model based on the probability relaxation method.According to the idea of global optimization,from the two aspects of the geometric similarity of the residential area itself and the object space similarity of its neighborhood,through the initialization of the probability matrix,the iteration of the probability matrix,and the selection of matching pairs,a matching model based on the probability relaxation method is proposed.Among them,the multi-factor weighting method is used to determine the initial probability matrix,the Delaunay triangulation network is constructed to index the neighborhood targets,the relative relationship between adjacent targets is calculated to calculate the support coefficient and compatibility coefficient,the iterative update of the probability matrix is completed,and the matching pair is finally selected.3)Design a comparative experiment for verification.Select the urban residential area map of a certain area in Shaanxi Province on the basic scale,and use the overlay matching method,the multi-factor weighting method,and the probability relaxation matching method distribution for comparison experiments.The results show that the probabilistic relaxation method has a good effect,can find the matching relationship between the new and old residential areas comprehensively and accurately,and effectively overcomes the problem of large position deviation between the new and old map targets.