Anisotropic Diffusion De-Speckling Based on Neutrosophic Theory

Effective speckle suppression and edge preservation are still core challenges in SAR image filtering. SAR images are inherently affected by speckle due to their coherent imaging mechanism. This speckle degrades image quality and complicates subsequent analysis. Traditional anisotropic diffusion (AD) filters have limitations in edge detection and fine detail preservation. To address the inherent speckle in synthetic-aperture radar (SAR) imagery, a novel de-speckling paradigm—Neutrosophic Theory Anisotropic Diffusion (NTAD)—is proposed. In NTAD, anisotropic diffusion governed by partial differential equations (PDE) is synergistically fused with the uncertainty-quantification formalism of neutrosophic theory. Indeterminacy, truth, and falsity memberships are explicitly encoded into the diffusion tensor, whereby the classical AD framework is endowed with enhanced robustness to SAR speckle while diagnostically relevant edge structures are simultaneously preserved. The proposed method improves edge localization accuracy while achieving superior speckle reduction and structural detail retention compared to conventional AD approaches. Experiments on three SAR image scenes demonstrate that the proposed NTAD method yields noticeably improved filtering performance.