欢迎来到天天文库
浏览记录
ID:46608455
大小:848.21 KB
页数:5页
时间:2019-11-26
《建筑物层次空间聚类方法研究》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、ComputerEngineeringandApplications计算机工程与应用建筑物层次空问聚类方法研究邓敏1,孙前虎1,文小岳2,徐枫3DENGMinl,SUNQianhul,WENXiaoyue2,XUFen931.中南大学测绘与国土信息工程系,长沙4100832.长沙市国土资源测绘院,长沙4100003.湖南师范大学资源与环境学院,长沙4100811.DepartmentofSurveyingandGeo-informatics,CentralSouthUniversity,Changsha41008
2、3,China2.SurveyingandMappingInstituteofChangshaLandBureau。Changsha410000,China3.CollegeofResource&EnvironmentalScience,HunanNormalUniversity,Changsha41008I,ChinaDENGMin。SUNQianhu.WENXiaoyue.eta1.Hierarchicalspatialclusteringofbuildings.ComputerEngineeringandA
3、p-pficafions,2011。47(28):120-123.Abstract:Spatialclusteringprovidesaneffectiveapproachforgeneralizationofresidentialareainautomatedcartographicgeneralization.BasedongraphtheoryandGestaltprinciple,ahierarchicalapproachisproposedinthispaper.Thisapproachcanbeuti
4、lizedtodiscoverthegraphicalstructureformedbybuildings,whichisobtainedwiththeconsiderationofshape,sizeandneighboringrelations.TheneighboringrelationsaredeterminedbyDelaunaytriangulation,whichisaqualitativecon-straintamongbuildings.Aweightedneighboringstructura
5、lgraphisobtainedbysettingvisualdistanceastheweightofthelinkingedgebetweenadjacentbuildings.TwolevelsofquantitativeconstraintsaredevelopedbyconsideringtheGestaltfac·tors,i.e.proximity,orientationandgeometryofbuildings.Oneistherotatingcalipersminimumdistance;th
6、eotheristhegeo—metricsimilaritymeasure.Throughexperimentsitisillustratedthattheresultsbythehierarchicalspatialclusteringproposedinthispaperareconsistentwithhumanperception.Keywords:spatialclustering;automatedcartographicgeneralization;Gestaltprinciples;hierar
7、chicalconstraints;neighboringstructuralgraph摘要:建筑物空间聚类是实现居民地地图自动综合的有效方法。基于图论和Gestalt原理,发展了一种层次的建筑物聚类方法。该方法可以深层次地挖掘建筑物图形的视觉特性,将面状地物信息充分合理地表达在聚类结果中.依据视觉感知原理,借助Dealaunay三角网构建方法,分析了地图上建筑物的自身形状特,l生和相互问的邻接关系,并依据建筑物间的可视区域均值距离建立了加权邻近结构图,确定了建筑物的邻近关系(定性约束)。根据Gestalt准则将
8、邻近性、方向性和几何特征等量化为旋转卡壳距离约束和几何相似度约束.通过实例验证了层次聚类方法得到更加符合人类认知的建筑物聚类结果。关键词:空间聚类;地图自动综合:Gestalt准则;层次约束:邻近结构图DOI:10.37788.issn.1002.8331.2011.28.033文章编号:1002.833l(2011)28.0120.04文献标i只码:A中图分类号:P20
此文档下载收益归作者所有