Segmentation Of Satellite Image Using Geodesic Active Contour

Résumé: In this paper we present the method of satellites images segmentation. The technique is based on the geodesic actives contours using geometric approach. We shown that a particular case of the classical model of Contours actives or “snakes” is amounts seeking a geodesic curve in Riemann space whose metric one is defined by the contents of the image, i.e to find a curve of minimal length in a certain space. By regarding geodesic active contour as level zero of an explicit function (“level set” approach developed by Osher et Sethian [1]), the search for this geodesic curve induced the solution of an partial derivative equation. So this intrinsic approach does not force initial contour to have same topology as the final curve. It is not either necessary to know the topology of the solution a priori (number of objects to segment). This diagram is applied to a satellite image and the results obtained show the effectiveness of the method suggested.

Publié dans la revue: Models & Optimisation and Mathematical Analysis Journal