Flow-Matching Guided Deep Unfolding for Hyperspectral Image Reconstruction

2025-10-02

AI Summary

Hyperspectral imaging (HSI) suffers from severe degradation and spectral detail loss due to high hardware costs and the ill-posed nature of reconstructing 3D hyperspectral cubes from compressive measurements. To address this, we propose FMU—a deep unrolling network integrating a flow-matching prior, marking the first application of flow matching to HSI reconstruction. FMU synergistically combines generative modeling with compressive sensing optimization, achieving both interpretability and strong representational capacity. We further introduce a novel mean-velocity loss to enforce global consistency of the learned velocity field, thereby enhancing reconstruction robustness and accuracy. Extensive experiments on both simulated and real-world datasets demonstrate that FMU significantly outperforms state-of-the-art methods in joint spatial-spectral recovery, particularly excelling in reconstructing fine-grained spectral structures and achieving new state-of-the-art performance.

Technology Category

• Computer Vision: Other Foundations of Computer Vision
• Machine Learning: Feature Construction/Reformulation
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Vertical and domain-specific search

Abstract

Hyperspectral imaging (HSI) provides rich spatial-spectral information but remains costly to acquire due to hardware limitations and the difficulty of reconstructing three-dimensional data from compressed measurements. Although compressive sensing systems such as CASSI improve efficiency, accurate reconstruction is still challenged by severe degradation and loss of fine spectral details. We propose the Flow-Matching-guided Unfolding network (FMU), which, to our knowledge, is the first to integrate flow matching into HSI reconstruction by embedding its generative prior within a deep unfolding framework. To further strengthen the learned dynamics, we introduce a mean velocity loss that enforces global consistency of the flow, leading to a more robust and accurate reconstruction. This hybrid design leverages the interpretability of optimization-based methods and the generative capacity of flow matching. Extensive experiments on both simulated and real datasets show that FMU significantly outperforms existing approaches in reconstruction quality. Code and models will be available at https://github.com/YiAi03/FMU.

Problem

• Reconstructing hyperspectral images from compressed measurements
• Addressing severe degradation and spectral detail loss
• Integrating flow matching with deep unfolding framework

Innovation

• Integrates flow matching into deep unfolding framework
• Introduces mean velocity loss for global consistency
• Combines optimization interpretability with generative capacity