Road extraction based on fuzzy logic and mathematical morphology from pan‐sharpened ikonos images

Abstract

Abstract Automatic extraction of geospatial features has been the subject of extensive research in the past three decades. Here, an approach based on fuzzy logic and mathematical morphology is proposed, to extract main road centrelines from pan‐sharpened IKONOS images. In the IKONOS images, a standard deviation of 10 grey levels has been measured for the road classes. In the proposed fuzzy logic system, just one arbitrary pixel (up to a maximum of 3 pixels) provides an adequate initial value. Road identification requires neither the numbers of the classes nor the corresponding mean values; then, using advanced morphological concepts, the road centreline is extracted. The method is applied to pan‐sharpened IKONOS images of urban, suburban and rural areas around the Pyramids in Egypt, and Rasht City and Kish Island in Iran. The extracted road centrelines have an average error of 0·504 pixel and root mean square error of 0·036 pixel. The method is more accurate at road intersections and on curves than on straight sections of road. The extracted road is then used as a direct input to a geographical information system (GIS).