Road segmentation from satellite aerial images by means of adaptive neighborhood mathematical morphology

In this paper, we present a methodology for segmenting the road from satellite aerial images using adaptive neighborhood mathematical morphology (ANMM) dealing with multiscale image processing. It is increasingly accepted that accurate image of the road networks enables the study of the entire or a congested part of the cities. This is possible with the help of an aerial image. There is a strong demand to automate acquisition and update of road date. The system used in this paper is a fully automatic that extracts road information using the concept of adaptive neighborhood techniques. A study is also conducted to find its efficiency compared to other road tracking methods using region competition/region segmentation techniques. The basic idea in this approach is to substitute the extrinsically defined - shape , fixed - size structuring elements generally used for morphological operators. The last ones should fit to the local multi scale features of the image, with respect to a selected criterion such as luminance, contrast, thickness and curvature. Our aim is to extract as many roads as possible independent of how wide and how sinuous changes of the roads.