GPU-Based Soil Parameter Parallel Inversion for PolSAR Imagery

Polarimetric Synthetic Aperture Radar (PolSAR) can obtain ground polarization information by transmitting and receiving polarized waves. The polarization information of the ground soil can be inverted to the moisture and roughness information by the empirical models. With the massive increase of PolSAR data, the demand for efficient processing is gradually growing. In this paper, a GPU based surface parameter parallel inversion method is proposed to solve this issue in quantitative remote sensing. This paper improves computational efficiency by using instruction set optimization, algorithm redundancy optimization, and fast numerical operations. The experimental results show that the method can realize approximately 100 times faster than the original serial version on CPU. If only the calculation part is considered, the method should achieve more than 1000 times acceleration.