Road Extraction Based on Image Features of Raster Urban Traffic Map

Based on the image features of color raster urban traffic map image, a novel algorithm for road networks extraction is proposed. The algorithm includes two key steps. Firstly, according to the dissimilar shape and color characteristics of the different geographical elements, a measure is established to recognize non-road regions' color, and then the other measure is set up, which may identify easily roads' color. On the basic of those, the urban map image segmentation is implemented. Secondly, two noise classification rules are brought forward which are integrated with some existing noises classification methods to classify noises for the intersected map image. The road networks are extracted when all of noises are classified accurately. Experimental results show that the proposed algorithm performs effectively for the color raster urban traffic map images.