Multi-stage near-field millimetre-wave phase processing for quantitative visualization of subsurface corrosion in 3LPE-coated metallic pipelines

Abstract Three-layer polyethylene (3LPE) coated metallic pipelines are widely used in oil and gas transportation, yet localized subsurface corrosion poses a major threat to pipeline safety. This study proposes a physics-guided multi-stage near-field millimetre-wave (NMMW) phase-processing framework for quantitative visualization of artificial corrosion beneath 3LPE coatings. This architecture coordinates successive processing stages to address the coupled interference mechanisms. A hybrid branch-cut iterative calibrated least-square (HBC-ICLS) method is developed for phase unwrapping of the original phase sequence. An improved Kernel density estimation (KDE)-based background removal and polarization-prior-guided tensor anisotropic diffusion filtering (PP-TADF) are presented to suppress strong background reflections and directional artifacts. In addition, independent component analysis (ICA) is applied for feature enhancement and dimensionality reduction from the processed phase sequence. Experimental validation demonstrates that the proposed framework can visualize artificial defects with a minimum lateral dimension of 4 mm, with maximum relative errors of 1.46% in localization and 9.8% in projected-area estimation. These results highlight the framework's capability for reliable quantitative imaging of early-stage corrosion beneath 3LPE coatings, thereby providing complementary support for pipeline integrity assessment.