A variational model for PolSAR data speckle reduction based on the Wishart distribution

Area of Science:

• Remote Sensing
• Signal Processing
• Computational Imaging

Background:

• Speckle noise in polarimetric synthetic aperture radar (PolSAR) data degrades image quality and hinders accurate analysis.
• Existing speckle reduction methods often struggle to balance noise suppression with the preservation of spatial details and polarimetric information.

Purpose of the Study:

• To develop a novel variational model for speckle reduction in PolSAR covariance and coherency matrices.
• To introduce the first variational approach for despeckling the entire PolSAR covariance or coherency matrix.
• To enhance the preservation of spatial resolution, edges, point scatterers, and polarimetric scattering characteristics.

Main Methods:

• A variational model based on the complex Wishart distribution and multichannel total variation (TV) regularization for complex-valued matrices.
• Derivation of the WisTV-C (covariance) and WisTV-T (coherency) models using maximum a posteriori estimation.
• A convex relaxation iterative algorithm employing variable splitting and alternating minimization to solve the nonconvex variational problem.

Main Results:

• The proposed WisTV model effectively reduces speckle in extended uniform areas of PolSAR data.
• Demonstrated superior preservation of spatial resolution, edges, and point scatterers compared to existing methods.
• Maintained the integrity of polarimetric scattering characteristics in despeckled PolSAR images.

Conclusions:

• The WisTV variational model offers a significant advancement in PolSAR speckle reduction.
• The method provides a robust solution for enhancing the quality and interpretability of PolSAR imagery.
• This approach sets a new benchmark for preserving detailed information in noisy PolSAR data.