数学形态学和模式识别在建筑物多边形化简中的应用  

The Application of Mathematical Morphology and Pattern Recognition to Building Polygon Simplification

作　　者：王辉连[1] 武芳[1] 张琳琳[1] 邓红艳[1] 

机构地区：[1]信息工程大学测绘学院,河南郑州450052

出　　处：《测绘学报》2005年第3期269-276,共8页Acta Geodaetica et Cartographica Sinica

基　　金：国家自然科学基金资助项目(40471115)

摘　　要：针对居民地图形化简的一个方面———建筑物多边形的化简,提出一种在与地图比例尺相关的动态栅格和矢量数据相结合的数据模型支持下,综合利用数学形态学和神经网络支持下的模式识别的化简方法。在VisualC++环境下实现基于此方法的系统并进行实验,实验结果说明此方法在保持街区的形态特征上效果明显。这种方法将制图综合知识融入图形化简操作之中,是自动制图综合智能化的一次新的尝试。The simplification of inhabited map area, especially simplifying the urban inhabited map area, is a difficult problem of automatic map generalization all along. There are many research results about this item in recent years, most of them are vector-based algorithms, and the others are raster-based. There is no perfect method which can make satisfactory generalization of inhabited map area yet. The main reason of this status is that it is difficult to get and use the semantic and spatial information of inhabited map area, In this paper, we present an approach to simplifying building polygon-one aspect of the simplification of inhabited map area, which is supported by a data model combining the raster and the vector model and adopt the methods of mathematical morphology and pattern recognition assisted by Neural Network. A system based on this method is carried out in the environment of Visual C＋ ＋, the results of test prove that the method is in evidence on preserving the shape characters of urban blocks. This approach, integrating the knowledge of map generalization into the simplification of graph, is an attempt on intellectualizing automatic map generalization.

关 键 词：自动制图综合 建筑物多边形化简 数学形态学 句法模式识别 FREEMAN链码 神经网络 

分 类 号：P208[天文地球—地图制图学与地理信息工程]