【摘要】 建筑物聚类是大比例尺地图自动制图综合中需要解决的关键问题。通过分析Gestalt原理的邻近性、相似性等,采用建筑物重心、建筑物间的距离、建筑物与邻近线状地物要素间位置关系等参数描述建筑物。本文提出的建筑物智能聚类方法包含两个连续的步骤:首先计算建筑物的描述参数,利用SOM网络的聚类能力,进行建筑物的初步聚类;然后,利用SOM竞争层行列扫描的方法,对初步聚类的建筑物类簇进行精确划分,获得满足建筑物聚类的全局和局部约束条件等制图要求的建筑物聚类群组。更多还原

【Abstract】 Building grouping is a key problem in automated map generalization.Principles of Gestalt theories,i.e.proximity,similarity,and common directions,are analyzed.Three parameters,i.e.centroid coordinates,minimum distance between buildings,and location relationship between buildings and roads,are selected for building description.An approach is presented for intelligent building grouping.The process of building grouping consists of two consecutive steps.Firstly,the three parameters are calculated and recorded.Based on the clustering characteristic of self-organizing map(SOM),buildings are clustered preliminarily.Secondly,a method of grouping the result of clusters based on SOM is applied.The SOM output layer is scanned row by row and column by column respectively.A further subdivision is accomplished to retrieve building groups complying with the global constraints of contextual features,the local constraints from Gestalt principles.更多还原