AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Unaligned Image Guided Denoising via Cross-modal Conditional Flow Matching

Runmin Zhang, Linshan Li, Si-Yuan Cao, Zhu Yu, Lingyu Zhu, Huaqi Zhang, Hui-liang Shen

NeurIPS 2026 · Image generation › Fast diffusion sampling · approach: Flow matching distillation · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes UGD-FM, a cross-modal conditional flow matching framework for denoising a target image guided by an unaligned reference image with large 2D misalignment.

Problem

Image denoising guided by a reference image when the guide is not spatially aligned with the noisy target and misalignment is large-range and two-dimensional, combined with severe target noise.

Approach

Formulates guided denoising as a progressive process built on conditional flow matching across modalities, handling large-range 2D misalignment rather than only small-baseline rectified pairs.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use flow matching instead of direct regression (one step) (ablation)UGD-FM · FANVID a=0.03 s=0.3, 1-stepPSNR: 30.27 -> 30.28No clear effect
run two solver steps instead of one in standard FM (ablation)UGD-FM · FANVIDPSNR: 30.28 -> 30.34Improved
add stage-focused time sampling (ablation)UGD-FM · FANVID, 2-stepPSNR: 30.34 -> 30.42Improved
add inference-consistent state construction (ablation)UGD-FM · FANVID, 2-stepPSNR: 30.42 -> 30.58Improved
add adaptive sparse guidance propagation instead of no attention (ablation)UGD-FM · FANVID, 2-stepPSNR: 29.10 -> 30.58Improved
use sparse propagation instead of window attention (ablation)UGD-FM · FANVID, 2-stepPSNR: 30.03 -> 30.58Improved
use global attention for guidance matching (ablation)UGD-FM · FANVIDPSNR: OOMHurt
add both perceptual consistency and flow consistency losses (ablation)UGD-FM · FANVID, 2-stepPSNR: 29.78 -> 30.58Improved
add flow consistency loss only (ablation)UGD-FM · FANVID, 2-stepPSNR: 29.78 -> 30.45Improved
run two-step rather than one-step inferenceUGD-FM (Restormer) · FANVID a=0.03 s=0.3PSNR: 30.29 -> 30.58Improved
use guided NIR denoising instead of single-image RestormerUGD-FM · MS2 a=0.03 s=0.3PSNR: 30.69 -> 32.82Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Experimental setup

Builds on: flow matching

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy