Geometric characteristics of Iraq’s raster topographic maps used for automatic updating the road network

This paper is devoted to the problem of road network extraction from raster image. The task of road network extraction is formulated in common view. The approach to the road map extraction has been proposed which can be applied for topographic map updating and is based on image clustering by k-means method and on application of scanning algorithm for extraction of road network fragments. Road map description is formed as set of linear fragments with knowing parameters. These linear fragments are created by merging of smaller parts. Experimental researches were implemented for maps of 10 Iraq cities. Experimental results show in average the extraction precision of 86% (in comparison with human expert).