Abstract: In line generalization, results depend very much on the characteristics of the line. For this reason it would be useful to obtain an automatic segmentation and enrichment of lines in order to apply to each section the best algorithm and the appropriate parameter. In this paper two factors of curves classification are discussed, which are directionality and sinuosity of curve, analyze the curve characteristic parameters with relation to these two dimensions and present a methodology for applying a line-classifying based on Bayesian method for a coastline segmentation task. The procedure is that 1） different sorts of curves are selected as training samples at first; 2） a moving window along the coastline is used to abstract curve parameters of curve segment within the window and to process these parameters using PCA method; 3） the statistical distribution functions of parameters of different sorts of curves are estimated and examined, and the curve segments within the window are classified base on Bayesian method; 4） these continuous curve units which belong to identical class are merged, and the points dividing different sorts of curves segment are regarded as the segmenting points of coastline. A test is performed over two coastline from a 1∶25 000 and 1∶1 000 000 scale mapwith a recommendation of the value of the parameters of the moving window, and segmentation results are reasonable and are consistent with human cognition.

Key words: Bayesian method, coastline segment, map generalization

摘要： 在地图的多尺度表达中，线要素在一定尺度下的表达结果很大程度依赖于自身的形态特征。因此，对线要素按其形态特征进行分段并对每一段实施不同的化简算法或者选用合适的参数进行化简是非常有意义的。讨论了影响曲线形态（方向性和弯曲性）的特征参量，在对特征参量进行主成分分析的基础上提出了一种基于贝叶斯方法进行曲线分类而最终实现海岸线分段的模型。其基本步骤是将一定大小的窗口以一定的步长在待分段的海岸线上移动，计算各个窗口内的曲线段特征参量并提取其主成分，基于随机函数从各个窗口中选取一定数量的样本，依靠专家识别样本的曲线类别，用概率假设的方法估计和检验各类曲线的各个主成分参数的统计分布规律，基于贝叶斯模型判别所有窗口内曲线段的类别，最后以不同曲线类别的分界点作为曲线的分段点。采用本模型对美国西北部一定区域1∶25万和我国南海1∶100万部分海岸线进行了实验，分段结果与人的认知基本相符。

关键词: 贝叶斯方法, 海岸线分段, 地图综合