基于格式塔原则的线性建筑群空间分布模式提取

Extraction of Spatial Distribution Model of Linear Building Group Based on Gestalt Principles

为了解决困扰地理信息系统学者的空间群目标提取难题,在街区矢量图中识别出线性的建筑群目标,笔者以Gestalt视觉识别原则作为度量准则,将线性的建筑群目标表示为马尔科夫链模型,利用置信传播算法进行求解。实验结果表明,与传统的聚类方法相比,该方法的可识别性有了较大的提高;与主流的机器学习方法相比,该方法对样本的依赖程度较低。

In this paper, in order to solve the puzzles of extraction of space group troubled plagued scholars of geography information system and recognize linear building group objects in vector diagram of block, taken Gestalt vision recognition principles as measure criterion, linear building group objects are shown as Markov chain. And, the authors solve the problems by using Belief Propagation algorithm. The experiment results shows that, compared with the traditional clustering methods, the identifiability of this method has been greatly improved; compared with the mainstream machine learning methods, the dependence severity of this method on sample is lower.