Land-cover classification using fused PolSAR and PolInSAR features

Due to the low information content of individual SAR images, single-band SAR data do not provide highly accurate land cover classification. However, in areas under risk where rapid land cover mapping is required, the advantages of SAR which include cloud penetration and day/night acquisition, are evident in comparison to optical data. The main research goal of this study is to fuse different frequency Polarimetric SAR (PolSAR) data as well as Polarimetric Interferometric (PolInSAR) data for land cover classification. Fusion techniques at two different levels are applied and combined in this study: the Logistic Regression (LR) as 'feature based fusion' method and the Neural-Network (NN) method for higher level fusion. Based on the results presented in this research we found that fused features from different SAR frequencies are complementary and adequate for land cover classification. Moreover, it has been found that PolInSAR features are complementary to the PolSAR information and essential for producing accurate land cover classification.