Retrieving urban areas on Google Earth images: application to towns of West Africa

Abstract

This article presents original stepwise image processing providing spatial delimitation of urban areas from Google Earth images in West Africa. In such images, converted to grey level images, urban areas are distinguished from their surroundings thanks to a specific textural aspect that may be retrieved by means of relevant morphological filters. Extraction of urban patterns consists of image analysis based on methods of mathematical morphology. The morphological filters used in the stepwise processing are chosen according to the spatial resolution, the visual quality of the original images and the environmental characteristics of the towns under study. Results are then compared with spatial data provided by a public geographical database.