Comparison of various clues of textures for the discrimination of urban fabric starting from the satellite imagery to high-resolution

Abstract

The extraction of urban areas on a satellite image starting from only spectral information is insufficient. It is necessary therefore that the methods of extracting information from these images perform better. That is possible while trying to exceed the simple spectral analysis, pixel by pixel, while utilizing the vicinity i.e. texture. Among the various approaches, most rely on texture parameters. The textural analysis can contribute and bring an invaluable help to the discrimination and the analysis of the theme built starting from the satellite image; while utilizing approaches which highlight the structure and the texture of built space. This is why the analysis of the urban areas which we study here is founded on an analysis of the texture of the image. We present in what follows a comparison between the various clues of textures for the analysis of urban fabric starting from the satellite image to high-resolution.