Score-Based Diffusion Models as Principled Priors for Inverse Imaging

2023/04/23 by Berthy T. Feng, Feng, Berthy T., Jamie Smith +9 · 38 citations
Mathematics · Medicine · Physics and Astronomy · #Artificial intelligence #Bayesian inference #Bayesian probability #Computer science #Deblurring #Image (mathematics) #Image processing #Image restoration #Inference #Inverse problem #Mathematics #Model Reduction and Neural Networks #Numerical methods in inverse problems #Posterior probability #Prior probability #Probabilistic logic #Radiomics and Machine Learning in Medical Imaging

paper · pdf · doi:10.48550/arxiv.2304.11751

published in arXiv (Cornell University) (Cornell University)

openalex publication_date 2023/04/23 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Priors are essential for reconstructing images from noisy and/or incomplete measurements. The choice of the prior determines both the quality and uncertainty of recovered images. We propose turning score-based diffusion models into principled image priors ("score-based priors") for analyzing a posterior of images given measurements. Previously, probabilistic priors were limited to handcrafted regularizers and simple distributions. In this work, we empirically validate the theoretically-proven probability function of a score-based diffusion model. We show how to sample from resulting posteriors by using this probability function for variational inference. Our results, including experiments on denoising, deblurring, and interferometric imaging, suggest that score-based priors enable principled inference with a sophisticated, data-driven image prior.

Cited by

• SURE Guided Posterior Sampling: Trajectory Correction for Diffusion-Based Inverse Problems
• PSI3D: Plug-and-Play 3D Stochastic Inference with Slice-wise Latent Diffusion Prior
• SFBD-OMNI: Bridge models for lossy measurement restoration with limited clean samples
• Explainable Learning Based Regularization of Inverse Problems
• DAPS++: Rethinking Diffusion Inverse Problems with Decoupled Posterior Annealing
• Latent Refinement via Flow Matching for Training-free Linear Inverse Problem Solving
• Blind Strong Gravitational Lensing Inversion: Joint Inference of Source and Lens Mass with Score-Based Models
• Deep generative priors for 3D brain analysis
• Briding Diffusion Posterior Sampling and Monte Carlo methods: a survey
• Coupled Data and Measurement Space Dynamics for Enhanced Diffusion Posterior Sampling
• Local MAP Sampling for Diffusion Models
• A Data-Driven Prism: Multi-View Source Separation with Diffusion Model Priors
• Position-Blind Ptychography: Viability of image reconstruction via data-driven variational inference
• Does FLUX Already Know How to Perform Physically Plausible Image Composition?
• Geometric Autoencoder Priors for Bayesian Inversion: Learn First Observe Later
• A Recovery Theory for Diffusion Priors: Deterministic Analysis of the Implicit Prior Algorithm
• NnD: Diffusion-based Generation of Physically-Nonnegative Objects
• Diffusion models for inverse problems
• QUTCC: Quantile Uncertainty Training and Conformal Calibration for Imaging Inverse Problems
• Improving Multislice Electron Ptychography with a Generative Prior
• Adaptive Mask-guided K-space Diffusion for Accelerated MRI Reconstruction
• Exact Conditional Score-Guided Generative Modeling for Amortized Inference in Uncertainty Quantification
• Zero-Shot Solving of Imaging Inverse Problems via Noise-Refined Likelihood Guided Diffusion Models
• Measurement-Aligned Sampling for Inverse Problem
• Supporting renewable energy planning and operation with data-driven high-resolution ensemble weather forecast
• Multiscale guidance of protein structure prediction with heterogeneous cryo-EM data
• IRBridge: Solving Image Restoration Bridge with Pre-trained Generative Diffusion Models
• EquiReg: Equivariance Regularized Diffusion for Inverse Problems
• Learning Single Index Models with Diffusion Priors
• Preconditioned Langevin Dynamics with Score-Based Generative Models for Infinite-Dimensional Linear Bayesian Inverse Problems
• Dual Ascent Diffusion for Inverse Problems
• LatentINDIGO: An INN-Guided Latent Diffusion Algorithm for Image Restoration
• How many measurements are enough? Bayesian recovery in inverse problems with general distributions
• Whitened Score Diffusion: A Structured Prior for Imaging Inverse Problems
• Good Things Come in Pairs: Paired Autoencoders for Inverse Problems
• Exact Posterior Score Estimation for Solving Linear Inverse Problems
• A neural operator view on U-Nets for inverse imaging problems
• Unifying and extending Diffusion Models through PDEs for solving Inverse Problems

Related

• Variational Bayesian Imaging with an Efficient Surrogate Score-based Prior
• Provably Robust Score-Based Diffusion Posterior Sampling for Plug-and-Play Image Reconstruction
• Plug-and-Play Priors as a Score-Based Method
• Measurement Score-Based Diffusion Model
• Whitened Score Diffusion: A Structured Prior for Imaging Inverse Problems