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

TL;DR

This paper introduces DU-BM3D, a hybrid neural network that unrolls the BM3D denoising algorithm into a trainable model by integrating a U-Net, combining interpretability with learnable flexibility for improved denoising performance.

Contribution

We propose a novel deep unfolded framework that replaces fixed filtering in BM3D with a trainable U-Net, enhancing denoising while maintaining interpretability.

Findings

DU-BM3D outperforms classic BM3D in PSNR and SSIM.

The method excels especially at high noise levels.

It generalizes well across different noise regimes.

Abstract

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.