Preprocessing of multifrequency SAR images: application to geological study of karstic formations (Le Larzac-France)

The usefulness of SAR images to detect geological features has been widely demonstrated by numerous studies. Nevertheless, when applied to SAR images, lithologies classification algorithms and pattern recognition tools intended for the extraction of features of interest are often unable to separate correctly actual geological features from artefacts due to speckle microtexture. The objective of this paper is twofold. The authors first demonstrate that appropriate preprocessing by an advanced tool like the gamma-gamma MAP features retaining an adaptive speckle filter makes the discrimination of lithologies both easier for photo-geologists and possible for automatic classifiers. Next, the combined use of X- and C-band data allow the authors to show how their different sensitivity enhances both discrimination of lithologies and identification of geological features