Uncertainty Quantification in HSI Reconstruction using Physics-Aware Diffusion Priors and Optics-Encoded Measurements

2025-11-23

AI Summary

Hyperspectral image (HSI) reconstruction from compressive measurements is a severely ill-posed inverse problem; existing data-driven methods often generate hallucinated artifacts due to insufficient spectral diversity in training data—particularly failing to model metamerism. This paper proposes HSDiff, a Bayesian inference framework that integrates a physics-informed diffusion prior with an optical encoding forward model to enable physically consistent and spectrally diverse sample generation. We innovatively design a regionalized metameric blackbody prior and a union-based spectral upsampling strategy to enhance prior diversity; further, we introduce the first fusion of unconditional pixel-wise diffusion priors with backward diffusion sampling for uncertainty-calibrated reconstruction. Extensive evaluation across multiple snapshot HSI systems demonstrates that effective spectral encoding substantially improves posterior plausibility, reconstruction diversity, and physical consistency.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Feature Construction/Reformulation

Application Category

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

Abstract

Hyperspectral image reconstruction from a compressed measurement is a highly ill-posed inverse problem. Current data-driven methods suffer from hallucination due to the lack of spectral diversity in existing hyperspectral image datasets, particularly when they are evaluated for the metamerism phenomenon. In this work, we formulate hyperspectral image (HSI) reconstruction as a Bayesian inference problem and propose a framework, HSDiff, that utilizes an unconditionally trained, pixel-level diffusion prior and posterior diffusion sampling to generate diverse HSI samples consistent with the measurements of various hyperspectral image formation models. We propose an enhanced metameric augmentation technique using region-based metameric black and partition-of-union spectral upsampling to expand training with physically valid metameric spectra, strengthening the prior diversity and improving uncertainty calibration. We utilize HSDiff to investigate how the studied forward models shape the posterior distribution and demonstrate that guiding with effective spectral encoding provides calibrated informative uncertainty compared to non-encoded models. Through the lens of the Bayesian framework, HSDiff offers a complete, high-performance method for uncertainty-aware HSI reconstruction. Our results also reiterate the significance of effective spectral encoding in snapshot hyperspectral imaging.

Problem

• Reconstructs hyperspectral images from compressed measurements using Bayesian inference
• Addresses hallucination issues through physics-aware diffusion priors and metameric augmentation
• Quantifies uncertainty in reconstruction using spectral encoding and posterior sampling

Innovation

• Physics-aware diffusion priors for uncertainty quantification
• Pixel-level posterior sampling with measurement consistency
• Metameric augmentation using spectral upsampling techniques