Abstract: According to the feature of the Digital Raster Graphic(DRG), a novel algorithm of extracting the road networks is proposed by recognizing the graphic annotation. The annotation information and roads are identified from the original DRG by using color cluster analysis. By making use of the template matching and color cluster analysis, the annotation districts are marked. Based on the features of roads and relationships between the roads and the annotation, some methods are put forward to connect the road segments. Experimental results show that the correct rate of this method is above 98%, and it can get good result of the road extraction and obtain the annotation information in the process.

Key words: road extraction, color clustering analysis, feature matching, annotation recognition, breakpoint connection

摘要： 根据数字栅格地图特征，提出一种通过识别地图标注来实现道路提取的方法。对原始数字栅格地图采用基于视觉一致性的色彩聚类分析，分离出所需等级道路及标注信息，结合色彩聚类分析与特征匹配精确识别各类标注体，确定标注区域，分析各类标注与被覆盖道路的关系，根据道路特征实现中断道路的连通。实验结果表明，该方法对标注的正确识别率达到98%以上，能实现不同等级道路的提取，并获取各类标注信息。

关键词: 道路提取, 色彩聚类分析, 特征匹配, 标注识别, 断点连接

CLC Number: 

• TP391.41