Study On The Calculation Methods Of Similarity For Spacial Scenes Consisted Of Vector Area Objects

Spatial similarity is defined of matching and ranking of space according to the specific content and scale of the space. It emphasizes the same degree between two spatial objects or two spatial scenes in one respect or more aspects,and it can help explain some of the phenomena and their surroundings. Query and retrieval based on the spatial similarity is closer to people’s way of thinking when making qualitative spatial thinking.A spatial scene consists of a set of geographic objects and spatial relationship between them and the spatial characteristics of other types. Research on spatial scenes similarity mainly in scenes feature extraction,spatial feature of similarity and similarity metric of spatial relationships. Spatial scenes similarity is the key to geographic information retrieval and multi-source geospatial data integration and updating, and research on measurements methods of spatial scenes similarity is of great significance for the intelligent geographic information retrieval mode in the future.This paper mainly focuses on the problem that the present research can only measure the spatial scenes similarity of which have an equal number of spatial objects.To solve this problem, we proposes a method to compute the scenes similarity of which have random object numbers. Firstly we extract the spatial scenes feature, then create scenes association graphs accroding to the spatial scenes feature.Finally we automatically calculate and matching spatial scenes set of database based on the scenes association graphs. According to its full extent, we divide the spatial scenes into exactly matching scenes, not exactly matching scenes. The existing models mostly add up the similarity values of all the objects and relationships between objects and then get the average value as the scenes similarity, but this way of calculation often give up of not matching exactly scenes which have a partial high matching degree. Aiming at this problem, this paper introduce spatial scenes objects weight and relations weight to perfect similarity measure model, then we can achieve the spatial scenes similarity measurement of any number of spatial objects.At last, this paper takes the vector data of a localized area in Xi’an city to verify the methed proposed in this paper and analyse and evaluate the matching results. The experiment results indicates that the method can effectively implete the similarity matching of spatial scenes in which are vector area objects. The method can take both completeness degree of spatial scenes and the sectional part of the scenes into consideration,and it can meet the needs of different search and query by adjusting the values of weight parameters.