SAR Image classification using textural modeling

It is recognized that texture is an important factor to discriminate forest types and other land cover in Synthetic Aperture Radar (SAR) images. This paper outlines an approach for texture modeling which is suitable for land identification using SAR images. The model is non parametric and is based on a likeness measure between neighborhoods. This study covers the problem of testing the mathematical descriptions of the model by using them to generate synthetic textures analogous with the source textures from which they were derived. This model has multiple applications in many image processing tasks such as texture segmentation, texture compression, inpainting, etc. We focus, here, on the practical application of terrain mapping.