Hyperspectral Intrinsic Decomposition: Joint Recovery of Reflectance and Photometric Components for Non-Lambertian Scenes

2026-07-28

AI Summary

This work addresses the limitations of existing hyperspectral intrinsic decomposition methods, which often rely on the Lambertian assumption and struggle with real-world non-Lambertian reflectance—particularly complex specular reflections—while frequently requiring auxiliary inputs or recovering only partial components. Building upon the dichromatic reflection model, the authors reformulate the joint recovery of four coupled reflectance and photometric components as the estimation of two spectral-spatial variables. They propose a dual-scale decomposition strategy: at the global scale, photometric-invariant edge priors preserve intrinsic boundaries; at the local scale, a specular-guided attention mechanism finely resolves specular-dominated regions. The study introduces CITE, the first real-world hyperspectral intrinsic decomposition dataset for non-Lambertian scenes, along with a physically plausible intrinsic image synthesizer (PISG), enabling blind, complete four-component decomposition without auxiliary information. Experiments demonstrate superior decomposition accuracy and robustness over state-of-the-art methods on CITE and multiple hyperspectral images.

Abstract

Hyperspectral intrinsic decomposition (HID) aims to disentangle material-related spectral properties and photometric effects in hyperspectral images (HSIs), which is essential for understanding real-world imaging processes and benefits a variety of downstream applications. Most existing HID studies have been developed under Lambertian or near-Lambertian assumptions. The few prior non-Lambertian efforts rely on simplified specular assumptions insufficient to handle diverse real-world specularity, and typically require auxiliary inputs or recover only a subset of the coupled reflectance and photometric components, hindering complete and blind decomposition. In this paper, we revisit the dichromatic reflection model (DRM) and develop a unified inversion paradigm that reformulates the recovery of four coupled reflectance and photometric components as the estimation of two spectral--spatial target variables. Building on this reformulation, we propose a dual-scale decomposition scheme to handle non-Lambertian effects with distinct spatial characteristics. At the global scale, photometrically invariant descriptors serve as edge priors for high-fidelity intrinsic boundary preservation; at the local scale, specularity-guided attention directs refinement with emphasis on specularity-dominated regions, including those affected by clipping distortion. To facilitate future research, we establish CITE, the first public real-world HID dataset for non-Lambertian objects, and develop a Physically-faithful Intrinsic Set Generator (PISG) for controllable data synthesis. Extensive ablation studies and experiments on the CITE and additional HSIs demonstrate the effectiveness of our method and its robustness across diverse scenes.

Problem

• Hyperspectral Intrinsic Decomposition
• Non-Lambertian
• Reflectance Recovery
• Photometric Components
• Specularity

Innovation

• hyperspectral intrinsic decomposition
• non-Lambertian
• dichromatic reflection model
• dual-scale decomposition
• specularity-guided attention