在建立GIS的过程中，从现有的各种比例尺的地形图上获得地理信息是必不可少的工作。街区式居民地是基 例尺地形图上最主要的地物要素类别之一，地在图二值图象的基础上，根据带晕线多边形的结构物征，利用一系列图象变换技术，首次成功地实现了街区式居民地的自动识别与提取。

It is indispensable to acquire geographic information from the existed topographical maps with various scales in the process of establishing geographic information system. The hatched polygons, which represent residential areas, are one of the most important sorts of map information. The strategies and algorithms of recognizing and extracting hatched polygons based on scanned map images are put forward and discussed. The RLS transformations in four directions for segmenting the hatched polygons are developed and realized. In addition, the algorithms for image transformations such as shrink, expansion, thinning, open graph deletion and hatched polygon recognition are also discussed in detail. Experiments show that recognition rate is about 96 percent with the methods developed in this paper.