第48卷第2期2023年4月Vol.48No.2Apr.2023测绘地理信息JournalofGeomatics一种不同比例尺面状居民地的概率松弛匹配算法刘洁1郭庆胜1王勇2陈恒11武汉大学资源与环境科学学院，湖北武汉，4300792中国测绘科学研究院，北京，100830AProbabilisticRelaxationMatchingAlgorithmforArealSettlementsatDifferentScalesLIUJie1GUOQingsheng1WANGYong2CHENHeng11SchoolofResourcesandEnvironmentScience，WuhanUniversity，Wuhan430079，China2ChineseAcademyofSurveyingandMapping，Beijing100830，China摘要：为了提高居民地面实体匹配的准确度，提出一种基于概率松弛模型的不同比例尺面状居民地匹配算法。首先，依据道路信息划分面状居民地数据，每次运行算法只处理一个道路网眼内的面状居民地；然后，扩展初始匹配模式，正反向计算居民地面实体之间和居民地面实体匹配到空两种匹配概率，并根据邻近匹配对的兼容性不断更新初始概率矩阵，直至相邻两次迭代矩阵内匹配概率变化量均小于某一阈值；最后，根据设定的选取规则选出1∶0、1∶1、1∶M和M∶N等4种匹配对。实验结果表明，本文所提方法可以有效准确地识别出4种匹配关系，明显减少了误匹配和漏匹配。关键词：面状居民地；实体匹配；道路网眼；概率松弛中图分类号：P208文献标志码：AAbstract：Inordertoimprovethematchingaccuracy，thispa‐perproposesamatchingalgorithmforarealsettlementsatdif‐ferentscalesbasedonprobabilisticrelaxationmodel.Thepro‐posedmethodfirstpartitionstwoarealsettlementsdatasetsin‐tosubsetsusingroadinformation，sothatateachrunthealgo‐rithmdealsonlywiththearealsettlementsinoneroadmesh.Afterthat，theinitialmatchingmodeisextendedtocalculatebidirectionalmatchingprobabilitybetweenarealsettlementsandbetweenarealsettlementtonull.Andthen，continuouslyupdatestheinitialprobabilitymatrixaccordingtothecompati‐bilityofadjacentmatchingpairsuntilthematchingprobabilityvariationsinmatrixoftwoconsecutiveiterationsarelessthanacertainthreshold.Finally，fourmatchingpairsof1∶0，1∶1，1∶MandM∶Nareselectedaccordingtotheselectionruleweset.Theresultsindicatethatthealgorithmproposedinthispa‐percaneffectivelyandaccuratelyidentifyfourmatchingrela‐tionshipsandsignificantlyreducethenumberofmismatchedpairsandmissedmat...
