Physics-Informed Untrained Learning for RGB-Guided Superresolution Single-Pixel Hyperspectral Imaging

2026-04-03

AI Summary

This work addresses the challenge of high-fidelity hyperspectral image reconstruction from single-pixel measurements under extremely low sampling rates (e.g., 6.25%), where the ill-posed inverse problem and scarcity of large-scale training data hinder existing deep learning approaches. The authors propose a novel end-to-end, physics-informed framework that requires no external training data, uniquely integrating an untrained neural network with the physical imaging model and leveraging an RGB image as a spatial-spectral prior. The method employs a three-stage physically constrained optimization strategy: initialization using grayscale priors derived from the RGB image, followed by an untrained hyperspectral recovery network (UHRNet), and a Transformer-based untrained super-resolution network (USRNet). This approach jointly optimizes hyperspectral reconstruction and super-resolution, achieving state-of-the-art performance in recovering 144-band hyperspectral cubes and demonstrating practical efficacy on a real single-pixel imaging system.

Abstract

Single-pixel imaging (SPI) offers a cost-effective route to hyperspectral acquisition but struggles to recover high-fidelity spatial and spectral details under extremely low sampling rates, a severely ill-posed inverse problem. While deep learning has shown potential, existing data-driven methods demand large-scale pretraining datasets that are often impractical in hyperspectral imaging. To overcome this limitation, we propose an end-to-end physics-informed framework that leverages untrained neural networks and RGB guidance for joint hyperspectral reconstruction and super-resolution without any external training data. The framework comprises three physically grounded stages: (1) a Regularized Least-Squares method with RGB-derived Grayscale Priors (LS-RGP) that initializes the solution by exploiting cross-modal structural correlations; (2) an Untrained Hyperspectral Recovery Network (UHRNet) that refines the reconstruction through measurement consistency and hybrid regularization; and (3) a Transformer-based Untrained Super-Resolution Network (USRNet) that upsamples the spatial resolution via cross-modal attention, transferring high-frequency details from the RGB guide. Extensive experiments on benchmark datasets demonstrate that our approach significantly surpasses state-of-the-art algorithms in both reconstruction accuracy and spectral fidelity. Moreover, a proof-of-concept experiment using a physical single-pixel imaging system validates the framework's practical applicability, successfully reconstructing a 144-band hyperspectral data cube at a mere 6.25% sampling rate. The proposed method thus provides a robust, data-efficient solution for computational hyperspectral imaging.

Problem

• single-pixel imaging
• hyperspectral imaging
• ill-posed inverse problem
• low sampling rate
• data-efficient reconstruction

Innovation

• Physics-Informed Learning
• Untrained Neural Networks
• RGB-Guided Super-Resolution
• Single-Pixel Hyperspectral Imaging
• Cross-Modal Attention