Constraint least-squares based method for pseudo quad-pol image reconstruction from compact data

Compact polarimetry is a hybrid polarity imaging mode. In this study, we propose a new algorithm based on least-squares (LS) error to estimate the quad-pol coherency elements. Different from the existing quad-pol reconstruction methods, the proposed method can be directly applied to the single look complex compact data, while all the other existing algorithms are based on the multi-look speckle reduced data. A model-based polarization state extrapolation model is derived, and then the system of nonlinear equations is optimized by the constraint LS estimation in a local window. Experiments are performed on C-band RADARSAT-2 data. In comparison with the iterative methods, the LS based method produces better overall reconstruction accuracy.