A Box Gradient‐Based Anisotropic Diffusion Framework for Polarimetric SAR Despeckling

Polarimetric synthetic aperture radar (PolSAR) images are inevitably corrupted by multiplicative speckle noise, which degrades image quality and subsequent interpretation tasks. This letter proposes a Box gradient‐driven anisotropic diffusion framework for PolSAR despeckling. By extending Box's test to the complex domain through the Pauli scattering vector, a statistically robust gradient is constructed to guide diffusion. The proposed method enables effective speckle suppression while preserving polarimetric structures, edges, and point targets. Experiments on real PolSAR datasets demonstrate that the proposed approach outperforms representative local and nonlocal filters in terms of noise reduction and structural preservation.