Research On Settlement Generalization Methods Considering Spatial Pattern And Road Networks

Map is one of the most effective tools for the description,analysis and transmission of spatiotemporal information.As one of the core contents of map making,multi-scale representation and so on,map generalization is a systematic and continuous scientific problem involving a wide range of technical difficulties.Generally,the idea of map generalization plays a vital role in spatial data integration,storage,transmission,representation,analysis,visualization and derivation process,and acts as basic research in the fields of spatiotemporal information acquiring,processing and representation.Collaborative map generalization,spatial pattern recognition and maintenance,m-n features mapping and relationship mapping cartographic generalization are much more difficult in map generalization and multi-scale representation.Thence,with the perspective of collaborative map generalization,this research aims to deeply excavate the key theories and technologies of settlement generalization considering spatial pattern and road networks as constraints.The main contents and innovations of this thesis are as the followings:(1)The research development of map generalization is reviewed and analyzed in four aspects of generalization knowledge category and acquisition,operator and algorithm,quality evaluation and process modeling and controlling.The basic theories and methods of settlement generalization are expounded from five aspects of principles,standards,indicators,demarcation size and quantity.Aiming at the selection quantity in multi-scale spatial database environment,an interpolated radical law is proposed.It can guarantee the quantity continuity of the results and the existing multi-scale database,and avoid the logic and quantity inconsistencies caused by normal radical law.(2)Multi-level collaborative map generalization.Taking settlement generalization as an example,the collaborative map generalization is studied in three levels of multi-feature,multi-operator and multi-algorithm.In this paper,multi-feature collaborative generalization refers to mesh based settlement generalization and point settlement selection considering adjacent road networks,road roundabout and corner;multi-operator generalization refers to typification which is regarded as a progressive and iterative process consisting of three basic operators including elimination,exaggeration and displacement;multi-algorithm generalization refers to mathematical morphology approach to settlement generalization based on Minkowski sum and multi-constrained point settlement selection.These studies above provide research probability for collaborative map generalization.(3)Vector domain mathematical morphology approach to settlement generalization.Traditional mathematical morphology is for raster data.In this paper,according to the connotation and nature of the Minkowski sum,extended morphology operators are put forwards for vector data.Through multi-level combination of Minkowski sum,a vector domain mathematical morphology approach is proposed for settlement generalization,which can keep the generalized settlement regular and squared.Taking the full advantage of vector data structure,the vector domain mathematical morphology can avoid the transformation between vector and raster and reduce additional computation.The proposed method uses polygonal structure element to detect,identify and generalize the input shape characteristics,and it breaks through the existing ideas of point-or polyline-based settlement generalization.(4)Lane-level road cluster recognition and typification algorithms.Based on the problem of polyline clustering analysis,a polyline cluster recognition algorithm is proposed.This paper firstly analyzes the concept and causes of lane cluster,then discusses the spatial constraints,and puts forwards the calculation method of moving distance.A lane-level road cluster mining algorithm with bidirectional region growing is proposed using distance and orientation.A typification method is put forwards to re-represent the lane-level road cluster.The polyline cluster recognition,re-representation and simplification will fill the related research fields.(5)Modeling and recognition of fuzzy spatial concepts such as spatial pattern.Linear pattern and grid pattern in buildings and road roundabout and road corner are common spatial concepts in map generalization,map making and related cartographic specifications.The linear and grid pattern recognition algorithms are proposed using Gestalt principle,computational geometry and graph theory.An adjacency length method is proposed to identify road roundabout and road corner,and a quantitative importance measurement is given.These are beneficial attempts to model,measure and identify spatial concepts,which will contribute to the generalization knowledge acquisition.(6)Building typification considering spatial pattern.Taking building typification as an example,the multi-operator level collaborative map generalization is studied.Through process identification,the typification operator is regarded as a progressive and iterative process consisting of three basic operators including elimination,exaggeration and displacement.Based on the roles and order,the three basic operators are effectively organized to achieve the function of typification,which has great advantages in the guarantee of initial position,spatial characteristics and spatial patterns including linear pattern and grid pattern.With this,a feasible research paradigm is provided for the multi-operator level collaborative map generalization,and also a new way is provided for m-n features mapping cartographic generalization.(7)Distance constrained Voronoi model.The normal Voronoi uses the rectangular bounding box of the input data as the boundary without the consideration of the spatial distribution characteristics of the input data.What's more,due to the boundary effect,the influence of the features at the boundary is mistakenly amplified.In this paper,a distance constrained Voronoi model is proposed,which can avoid the spatial conflict,avoid excessive temporary solidification to some extent,and optimize the ability of spatial distribution preserving.The distance constrained Voronoi diagram model has great significations for the data of irregular shaped or group distributed.(8)Point settlement selection based on multiple constraints.According to the principles of settlement selection,an improved multi-constrained generic selection strategy is proposed and applied to practical projects,which takes several constraints into account including multi-scale quantity consistency,semantic rules,spatial distribution,spatial conflict,adjacent roads,road roundabouts and corners.This strategy integrates the advantages of existing algorithms,and realizes the multi-algorithm level collaborative settlement selection,which provides feasible cases for multi-algorithm level collaborative map generalization.