Abstract: This paper studies a method on the recognition and extraction of road network from a color raster image of city traffic map. Based on the mathematical, model of the map image, according to the object color features on the color map image, we first make a normalization processing to the image. Then the noise characteristics on the color raster map are analyzed. By integrating the noise characteristics with the gray-scale morphology's operator, an effective method is proposed in the paper. This method can wipe out the noises on the color raster map image. Thereby, the recognition and extraction of the road network layer from the color raster city traffic map image can be implemented. An example is also given in the paper to show the effectiveness of the recognition and extraction process.