Research On The Spatial-scene Similarity Retrieval Based On Spatial Relations

With the rapid development of the spatial data acquisition technology,the scale of spatial data is explosively growing,but the utilization ratio of the spatial data is very low.To improve the utilization ratio of the spatial data,the shape feature,topological relation,and directional relation of the spatial data should be studied.By means of the description and retrieval of the spatial object shape feature,description and retrieval of the spatial topological relation,and description and retrieval of the directional spatial relation,the accurate and effective spatial-scene similarity retrieval is achieved.In order to solve the problem that the existing shape description methods based on the boundary are sensitive to the deformation,and that the existing shape description methods based on the region are easy to ignore the details of the shape,the omnidirectional feature description method of the shape has been studied in this dissertation,which can be used to describe and retrieve the spatial object shape feature.This method can combine the boundary information and region information of the shape,and combine the global features and local features of the shape in order to improve the retrieval accuracy of the shape.This method is invariant to translation,scaling,rotation,and has the strong ability to describe and recognize the deformed shape.This method can perform the similarity retrieval of the spatial object shape with high precision,provide the basis for the analysis of the spatial relations,and be the foundation of the spatial-scene similarity retrieval.In order to solve the problem that the existing topological relation description methods are not strong enough to describe the details,and that the unified mechanism of the topological similarity measurement is missing,the quantitative description method of the spatial topological relation has been studied in this dissertation,which can be used to describe and retrieve the spatial topological relation.This method can measure the topological relation quantitatively,relax the constraint of the topological relation,and unify the similarities of the quantitative measurement and constraint relaxation in order to improve the recognition ability of the topological relation and expand the retrieval scope of the topological relation.This method is invariant to translation,scaling,and rotation.This method can perform the similarity retrieval of the spatial topological relation with high precision and provide the basis for the spatial-scene similarity retrieval.In order to solve the problem that the evaluations of the existing directional spatial relation description methods are sometimes different from the human space perception,the position sensing description method of the directional spatial relation has been studied in this dissertation,which can be used to describe and retrieve the directional spatial relation.This method is based on the concept of the spatial position perception,and combined with the position sensing parameters.By means of the histogram of position sensing forces,this method can quantitatively describe the relative position of the argument object with regard to the reference object in the exact direction and inexact direction,and the directional spatial relation of the reference object surrounding the argument object.This method can simulate the human cognition for the description of the directional spatial relation.This method is invariant to translation,scaling,and rotation.This method can perform the similarity retrieval of the directional spatial relation with high precision and provide the basis for the spatial-scene similarity retrieval.In order to solve the problem that the existing spatial data retrieval methods can't consider the shape features,and can't adaptively calculate the similarity of the spatial scene,the spatial-scene similarity retrieval method has been studied in this dissertation.This method is based on the shape similarity matching,topological relation similarity matching and directional spatial relation similarity matching.This method can retrieve all the database scenes completely matching or partially matching with the sample scene,and adaptively determinate the importance of the shape feature,directional spatial relation and topological relation.This method can unify the similarities of the shape feature,topological relation and directional spatial relation for the similarity of the spatial scene,and provide the rational similarity scores for completely matching scenes and partially matching scenes.The spatial-scene similarity retrieval has high accuracy and is consistent with the human retrieval habit and spatial cognition.In this dissertation,the spatial-scene similarity retrieval method can unify the similarity matchings of the shape feature,topological relation,directional spatial relation and metric information by means of the omnidirectional feature description method of the shape,quantitative description method of the topological relation,and position sensing description method of the directional relation.This method has high retrieval accuracy of the spatial scene and is an effective and accurate spatial data retrieval scheme.