地形图点状符号的自动提取和识别  

Automatic Extracting and Identifying of Point-shaped Symbols on Topographic Map

作　　者：陈飞[1] 王秀芳[2] 王坤[1] 农宇[1] 

机构地区：[1]武汉大学资源与环境科学学院,武汉430079 [2]武汉大学遥感信息与工程学院,武汉430079

出　　处：《自动化学报》2007年第10期1074-1080,共7页Acta Automatica Sinica

摘　　要：针对扫描地形图的点状符号,提出了一种基于距离变换进行提取和识别的新方法.首先优化分版图的提取结果,根据弯曲密度和线划密度过滤掉非点状符号的线划,利用加壳变换和蜕皮变换进行点状符号的图文分离.然后根据加权距离函数来度量符号的全局特征,由骨架线的一致性米度量符号的几何形状特征,并结合这两个特征来识别点状符号.最后,提出了错切文字注记和线状要素注记的识别方法.This paper presents a new method of extracting and identifying point-shaped symbols of scanned topographic maps based on distance transformation. First, we optimize the extracting results of element maps, and filter the lines which do not meet the features of point symbols according to camber density and line density, and detach the symbols from the map based on shelling transformation and molting transformation. Next, we measure the whole features of symbols by a weighed distance function and the geometry shape feature by the consistence of symbol skeleton, and identify symbols by these two features. The algorithm can identify the shear transform Chinese character and annotations of line symbols.

关 键 词：地形图 点状符号 距离变换 线划弯曲度 符号骨架线 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]