Entropy based reconstruction technique for analysis of hybrid-polarimetric SAR data

What is it about?

The analysis of hybrid-polarimetric (pol) data can be performed using any of the well established full-pol decomposition techniques through reconstruction of pseudo quad-pol data. This reconstruction requires estimation of cross-polarization information from the hybrid-pol measurements. In this paper, we established a direct relationship to estimate from hybrid-pol entropy. This proposed reconstruction approach overcomes the limitations related to the iterative based procedure implemented in the conventional Souyris’ reconstruction method. The proposed method also circumvents a non-linear co- and cross-pol linking used in Souyris’ method that holds only for volume scattering dominated areas. The comparative analysis over four different datasets indicate that the proposed method can perform better for the application of landcover classification, in comparison with conventional Souyris’ as well as recently reported Espeseth’s reconstruction methods.

Featured Image

Why is it important?

The advantage with the proposed method is that it overcomes the limitations associated with Souyris’ reconstruction method. Validation over four different datasets indicate that the proposed technique outperforms the Souyris’ and Espeseth’s Eigenvalues-based techniques in single- and double-bounce dominating regions.