AI Paper Atlas

Research map / Image generation / Image restoration

Hierarchical Integration Diffusion Model for Realistic Image Deblurring

Zheng Chen, Yulun Zhang, Ding Liu, Bin Xia, Jinjin Gu, Linghe Kong, Xin Yuan

NeurIPS 2023 · Image generation › Image restoration · approach: Diffusion priors for denoising and deblurring · 29 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add diffusion-generated prior with hierarchical integration to Restormer (ablation)Restormer · GoPro (ablation setting)PSNR (dB): 31.96 -> 32.24Improved
use multi-scale prior instead of single-scale prior (ablation)HI-Diff · GoPro (ablation setting)PSNR (dB): 32.00 -> 32.24Improved
jointly train diffusion model and Transformer in stage two (ablation)HI-Diff · GoPro (ablation setting)PSNR (dB): 30.73 -> 32.24Improved
single-scale prior guide versus no prior (ablation)Restormer · GoPro (ablation setting)PSNR (dB): 31.96 -> 32.00No clear effect
increase diffusion iterations from 1 to 8 (ablation)HI-Diff · GoProPSNR (dB)Improved
increase diffusion iterations beyond 8 (to 16, 32) (ablation)HI-Diff · GoProPSNR (dB)No clear effect
replace Restormer with HI-DiffRestormer · GoProPSNR (dB): 32.92 -> 33.33Improved
replace Restormer with HI-DiffRestormer · HIDEPSNR (dB): 31.22 -> 31.46Improved
replace Stripformer with HI-Diff trained on RealBlurStripformer · RealBlur-J (trained on RealBlur)PSNR (dB): 32.48 -> 33.70Improved
replace Stripformer with HI-Diff trained on RealBlurStripformer · RealBlur-R (trained on RealBlur)PSNR (dB): 39.84 -> 41.01Improved
add diffusion prior to Restormer (ablation)Restormer · GoPro (ablation setting)FLOPs (G): 117.25 -> 125.47Hurt

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.

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