$\mathrm{S}^3$-PnP: Plug-and-Play Denoising for Scientific Imaging with Self-Supervised Score Priors

Key points

• Modern image restoration methods increasingly rely on large-scale supervised training, yet clean targets are often prohibitively difficult to obtain in scientific imaging modalities such as hyperspectral imaging (HSI) and magnetic resonance imaging (MRI).1
• Plug-and-play ADMM (PnP-ADMM) with pretrained diffusion priors offers a promising alternative, but the learned prior distribution can mismatch both the target scientific image distribution and the distribution of ADMM denoiser inputs induced by dual feedback.1
• We further analyze the convergence and reconstruction error of the corrected ADMM algorithm assuming an ideal score and quadratic data fidelity.1
• Experiments on reference-guided HSI restoration, single-image HSI denoising, and MRI denoising demonstrate consistent improvements over competing methods.1

Sources (1)

• [1]$\mathrm{S}^3$-PnP: Plug-and-Play Denoising for Scientific Imaging with Self-Supervised Score Priors
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 13, 02:26 PM

  To mitigate these mismatches, we propose $S^3$-PnP, a self-supervised plug-and-play restoration framework that learns a score prior directly from observation-derived ADMM intermediate states, without clean training targets or externally pretrained weights.

  Modern image restoration methods increasingly rely on large-scale supervised training, yet clean targets are often prohibitively difficult to obtain in scientific imaging modalities such as hyperspectral imaging (HSI) and magnetic resonance imaging (MRI).

Extractive summary: sentences quoted from the sources.