Interferometric Synthetic Aperture Radar Coherence Histogram Analysis for Land Cover Classification

SAR interferometry has a large potential for topographic and geophysical parameters extraction algorithms. The interferometric coherence, derived from the cross-correlation of two complex coregistered SAR images, gives additional information complementary to information contained in the amplitude value of the backscattering coefficient. However, the interferometric phase is considered with less importance as it has no effective relationship with different land covers on the image. Applications of the interferometric SAR (InSAR) coherence for land use classification is discussed. We proposed an original land cover classification methodology based on the InSAR coherence histogram analysis. It consists on characterizing each land surface categories (lake, bare soil, urban...) with a signature formed with a probability density function (pdf) computed from the corresponding InSAR coherence class. Both qualitative and quantitative analysis are presented and discussed