Research On Theory And Methodology Of Road And Settlement Fusion Considering Scale Variation

Vector spatial data is an important carrier of human social information.Currently,various departments have produced vector spatial data of different scales,which have different levels of detail and differ in geometry,attributes and spatial relationships,bringing difficulties to the sharing,integration and application of spatial data.Multi-scale vector spatial data fusion is the key to solving the above problems,which is mainly to integrate and unify the spatial data of different scales.After spatial data matching,cartographic generalization and spatial relationship processing,the data is mutually confirmed,complemented and related with each other in attributes and geometric positions,so as to achieve the goal of integration and enhancement of spatial information.This dissertation mainly studies the theory and methodology of road and settlement fusion considering scale variation.The main tasks completed and the results obtained are as follows:1.The research background and significance are analyzed combined with actual requirements.This dissertation also summarizes the general research progress of spatial data fusion and the current research status of road and settlement fusion,and points out the shortcomings of the current research.Finally,this dissertation proposes research objectives,content,and technology route etc.2.From the aspects of spatial cognition and multi-scale representation,the basic connotation and characteristics of scale are expounded.The sources of differences and manifestation of multiscale vector spatial data are also deeply analyzed.The basic problems of multi-scale vector spatial data fusion are discussed.This dissertation also clarifies the specific concepts and research categories,and defines the specific classification and formal definitions.The specific research content and technical framework are proposed.Finally,the common descriptive features of road and settlement are introduced.3.The multi-scale road fusion method based on hierarchy is studied.An overpass recognition method based on target detection Faster R-CNN model is proposed.By using the convolutional neural network to learn the deep structure features of the overpass samples,the automatic recognition and accurate positioning of the overpass is realized.Aiming at the problem that the existing methods only consider the static characteristics of road network,a hybrid multi-feature selection method combining with trajectory data is proposed.The dual graph is constructed to describe the structural relationship of road network.And the Voronoi Diagram is used to divide and constrain the density of road network,which ensures the rationality of the selected results.A multi-scale road network matching method considering hierarchical constraints is proposed,which divides the road network into different levels from the perspective of spatial cognition,making full use of the spatial structure of the road network and constraining the matching process with the hierarchical structure.The method can be applied to road network matching of different scales.4.The method of multi-scale settlement fusion considering multi-features is studied.The Delaunay triangulation network is constructed.And the complex matching entities at different scales are divided into closed domains through spatial neighborhood clusters to obtain a set of many-to-many matching candidate elements.And on this basis,a multi-scale residential area matching method that takes multi-features into account is proposed.Considering the influence of self-features and spatial neighborhood features on the matching results,a spatial similarity measure model is constructed.Then,the final matching result is obtained.A Voronoi diagram point cluster selection method that takes multi-features constraints into account is proposed,which combines the spatial distribution,topology,density,thematic attributes,and the association with road elements of point cluster elements.The distance constraints are also applied to deal with spatial conflicts,which can maintain the consistency of spatial characteristics before and after point cluster selection.5.The consistency processing method of multi-scale road and settlement is studied.The influence of dimension reduction,simplification and mergence on multi-scale road and settlement spatial relation is analyzed.This dissertation introduces the description method of topological relation based on nine-intersection model,and divides the line-polygon topological relation into multiple levels.According to the line-polygon and line-point conflicts in map space and the linepolygon and line-polygon conflicts in vector space at different scales,the corresponding relations of line-polygon and line-point consistency,and the measuring relations of line-polygon and linepolygon consistency are established respectively.The method of spatial conflict detection is put forward,with the related experimental verification and analysis being carried out.A spatial constraint-oriented road and residential displacement processing method is proposed.This method uses the minimum spanning tree to establish the relationship between the buildings with the road mesh as the unit,shifts and zooms the buildings to effectively maintain the spatial distribution pattern of the buildings.6.A prototype system for fusion processing of multi-scale road and settlement is developed.The main functions of the system and the overview of experimental data are introduced.And the key technical methods proposed in the dissertation are experimentally verified.