Deep Unfolded BM3D: Unrolling Non-local Collaborative Filtering into a Trainable Neural Network

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Basim, Kerem, Unal, Mehmet Ozan, Ertas, Metin, Yildirim, Isa
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918203407466496
author Basim, Kerem
Unal, Mehmet Ozan
Ertas, Metin
Yildirim, Isa
author_facet Basim, Kerem
Unal, Mehmet Ozan
Ertas, Metin
Yildirim, Isa
contents Block-Matching and 3D Filtering (BM3D) exploits non-local self-similarity priors for denoising but relies on fixed parameters. Deep models such as U-Net are more flexible but often lack interpretability and fail to generalize across noise regimes. In this study, we propose Deep Unfolded BM3D (DU-BM3D), a hybrid framework that unrolls BM3D into a trainable architecture by replacing its fixed collaborative filtering with a learnable U-Net denoiser. This preserves BM3D's non-local structural prior while enabling end-to-end optimization. We evaluate DU-BM3D on low-dose CT (LDCT) denoising and show that it outperforms classic BM3D and standalone U-Net across simulated LDCT at different noise levels, yielding higher PSNR and SSIM, especially in high-noise conditions.
format Preprint
id arxiv_https___arxiv_org_abs_2511_12248
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Deep Unfolded BM3D: Unrolling Non-local Collaborative Filtering into a Trainable Neural Network
Basim, Kerem
Unal, Mehmet Ozan
Ertas, Metin
Yildirim, Isa
Image and Video Processing
Artificial Intelligence
Computer Vision and Pattern Recognition
Block-Matching and 3D Filtering (BM3D) exploits non-local self-similarity priors for denoising but relies on fixed parameters. Deep models such as U-Net are more flexible but often lack interpretability and fail to generalize across noise regimes. In this study, we propose Deep Unfolded BM3D (DU-BM3D), a hybrid framework that unrolls BM3D into a trainable architecture by replacing its fixed collaborative filtering with a learnable U-Net denoiser. This preserves BM3D's non-local structural prior while enabling end-to-end optimization. We evaluate DU-BM3D on low-dose CT (LDCT) denoising and show that it outperforms classic BM3D and standalone U-Net across simulated LDCT at different noise levels, yielding higher PSNR and SSIM, especially in high-noise conditions.
title Deep Unfolded BM3D: Unrolling Non-local Collaborative Filtering into a Trainable Neural Network
topic Image and Video Processing
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2511.12248