A Three-Component Polarimetric Target Decomposition Algorithm for Grasslands

Previous polarimetric target decomposition methods have been widely used in forests and buildings and have achieved good results. However, there is a lack of corresponding modeling for the study of grassland scattering characteristics. To improve the accuracy of grassland target decomposition, this paper fully considers the characteristics of grassland vegetation and abstracts the grassland vegetation canopy into a shape-adaptive random quasi-ellipsoid particle cloud. The random state of grassland vegetation is simulated from the three dimensions of spin angle, tilt angle, and rotation angle, and the shape of grassland vegetation is adaptively simulated by the anisotropy degree A to generalize the applicability of the model so that the particle cloud is more in line with the overall state of grassland vegetation. To better deal with unnatural distributed scatterers, that is, considering the correlation between co-polarization and cross-polarization, a helix scattering mechanism is introduced. The derived volume scattering model is constructed with the helix scattering model and the Freeman two-component (FTC) ground scattering model to form a novel three-component scattering model suitable for grasslands, and a polarimetric target decomposition algorithm is proposed based on this model. Experimental comparisons with several other excellent polarimetric target decomposition algorithms are carried out on X-band Cosmos-SkyMed, C-band GF-3, and L-band AfriSAR fully polarimetric data. The experimental results show that this paper's algorithm performs better than other algorithms for decomposition effect on different data, and is more reflective of the scattering characteristics of different grassland vegetation regions.