Spectral Compressive Imaging via Chromaticity-Intensity Decomposition

2025-09-20

AI Summary

CASSI-based hyperspectral image (HSI) reconstruction faces two fundamental challenges: spatial-spectral information entanglement and illumination dependence, leading to ill-posed recovery and illumination-sensitive reflectance estimation. To address these, this work proposes a novel chroma-luma decomposition framework—the first to introduce this principle into compressive spectral reconstruction—disentangling the HSI into a smooth luma component and a detail-rich chroma cube for illumination-robust reflectance estimation. Leveraging a dual-camera CASSI architecture, we design an unrolled hybrid spatio-temporal Transformer network that jointly models chroma sparsity and degradation characteristics. We further incorporate a degradation-aware module and a spatially adaptive noise estimation mechanism to accurately characterize anisotropic measurement noise. Extensive experiments on both synthetic and real-world datasets demonstrate significant improvements in spectral fidelity and chroma consistency, consistently outperforming state-of-the-art methods across all quantitative and qualitative metrics.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization
• Machine Learning: Feature Construction/Reformulation

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work

Abstract

In coded aperture snapshot spectral imaging (CASSI), the captured measurement entangles spatial and spectral information, posing a severely ill-posed inverse problem for hyperspectral images (HSIs) reconstruction. Moreover, the captured radiance inherently depends on scene illumination, making it difficult to recover the intrinsic spectral reflectance that remains invariant to lighting conditions. To address these challenges, we propose a chromaticity-intensity decomposition framework, which disentangles an HSI into a spatially smooth intensity map and a spectrally variant chromaticity cube. The chromaticity encodes lighting-invariant reflectance, enriched with high-frequency spatial details and local spectral sparsity. Building on this decomposition, we develop CIDNet, a Chromaticity-Intensity Decomposition unfolding network within a dual-camera CASSI system. CIDNet integrates a hybrid spatial-spectral Transformer tailored to reconstruct fine-grained and sparse spectral chromaticity and a degradation-aware, spatially-adaptive noise estimation module that captures anisotropic noise across iterative stages. Extensive experiments on both synthetic and real-world CASSI datasets demonstrate that our method achieves superior performance in both spectral and chromaticity fidelity. Code and models will be publicly available.

Problem

• Reconstructing hyperspectral images from severely ill-posed CASSI measurements
• Recovering intrinsic spectral reflectance invariant to lighting conditions
• Disentangling spatial-spectral information entangled in coded aperture snapshots

Innovation

• Chromaticity-intensity decomposition for reflectance separation
• Hybrid spatial-spectral Transformer for chromaticity reconstruction
• Degradation-aware noise estimation module for anisotropic noise