Classification of X-band high resolution SAR images over urban areas: Markovian segmentation using Mellin transform

This paper presents a classification method devoted to high-resolution SAR amplitude images over dense urban areas. The classification method is based on a statistical approach involving a Markovian segmentation process. Our contribution is the introduction of Fisher distributions to fit urban area statistics at best. This implies the use of second kind statistics at best. This implies the use of second kind statistics built upon Mellin transform. The estimation process and the Markovian process are described here. Our goal is both to explain how these distributions can be used in a classification process and to discuss some results. The theory is explained and its validity regarding our data is presented. Then the discussion focuses on the application of this theory to real data.