RetiDiff: Stable Underwater Image Color Reconstruction Based on Retinex and Diffusion Distillation

Simple item page
dc.contributor.authorQiu, Wenyao
dc.contributor.authorZhou, Zhuang
dc.contributor.authorZhang, Xin
dc.contributor.authorChen, Jiayi
dc.contributor.authorZhou, Shiping
dc.contributor.authorTao, Ran
dc.contributor.editorMusialski, Przemyslaw
dc.contributor.editorLim, Isaak
dc.date.accessioned2026-04-20T08:01:38Z
dc.date.available2026-04-20T08:01:38Z
dc.date.issued2026
dc.description.abstractUnderwater image color reconstruction remains challenging due to wavelength-dependent light absorption and scattering that cause severe color casts and visibility degradation. We propose RetiDiff, a Retinex-guided diffusion distillation framework that couples a physics-aware diffusion prior with a lightweight Retinex-based UNet for stable, single-pass color restoration. A conditional Diffusion Transformer (DiT), pretrained on physics-guided underwater–terrestrial pairs, is frozen and distilled via Score Distillation Sampling (SDS) into a Retinex-UNet that predicts reflectance R and illumination L. This distillation transfers domain-agnostic color priors while mitigating cross-domain feature entanglement and avoiding iterative diffusion. To further suppress artifacts from imperfect Retinex separation, an Inter-Component Residual (ICR) regularization penalizes cross-component correlation and gradient co-occurrence, reducing halos, ghosting, and color drift while preserving structural fidelity. Extensive experiments on UIEB, LSUI, and TEST-U45 demonstrate state-of-the-art perceptual quality and LAB-space fidelity, with RetiDiff achieving comparable or superior performance to diffusion-based baselines while requiring far fewer parameters, lower FLOPs, and an order-of-magnitude faster inference.
dc.description.sectionheadersAppearance, Imaging & Tools
dc.description.seriesinformationEurographics 2026 - Short Papers
dc.identifier.doi10.2312/egs.20261008
dc.identifier.isbn978-3-03868-299-8
dc.identifier.issn2309-5059
dc.identifier.pages4 pages
dc.identifier.urihttps://diglib.eg.org/handle/10.2312/egs20261008
dc.identifier.urihttps://diglib.eg.org/handle/10.2312/egs20261008
dc.publisherThe Eurographics Association
dc.rightsCC-BY-4.0
dc.rights.urihttps://creativecommons.org/licenses/by/4.0/
dc.subjectArtificial Intelligence
dc.subjectComputer Vision
dc.subjectImage Reconstruction
dc.titleRetiDiff: Stable Underwater Image Color Reconstruction Based on Retinex and Diffusion Distillation
Files
Original bundle
Now showing 1 - 2 of 2
Thumbnail Image
Name:
egs20261008.pdf
Size:
3.91 MB
Format:
Adobe Portable Document Format
Download
Thumbnail Image
Name:
short1007_mm_crc1.pdf
Size:
7.34 MB
Format:
Adobe Portable Document Format
Download
Collections
EG 2026 - Short Papers