面向对象分类分析提取地籍图地块边界  

Lot-edge extraction from cadastral map using object-oriented classification

作　　者：谭衢霖[1] Aldred Andrew 

机构地区：[1]北京交通大学土木建筑工程学院,北京100044 [2]西安大略大学地理系

出　　处：《北京交通大学学报》2013年第4期134-138,共5页JOURNAL OF BEIJING JIAOTONG UNIVERSITY

基　　金：国家自然科学基金资助项目(51078020);中央高校科研业务费专项资金资助(2012JBM078)

摘　　要：提出了一种基于面向对象分类分析的地籍图地块边界提取方法.该方法通过对纸质地籍图的扫描栅格图像进行分割,形成多个待分类的图像对象(区域).结合人工智能专家知识构建模糊逻辑分类器,从形成的多个图像对象中逐层分类提取出地块对象.最后,经矢量化和综合概括处理,最终获得矢量化的地籍图地块边界要素.A scheme based on object-oriented classification analysis was developed to extract lot-edge from scanned cadastral map in the paper. Firstly, segmentation was performed to form image objects or regions in the digital raster map. Secondly, four knowledge rules incorporated into a fuasy logic classifier were constructed for classifying lot and non-lot image obiects. Finally, lot polygons were extracted by vectorization and generalization.

关 键 词：地籍图 面向对象 图像分割 分类 知识规则 

分 类 号：P208[天文地球—地图制图学与地理信息工程]