Constructing an Interpretable Deep Denoiser by Unrolling Graph Laplacian Regularizer

TL;DR

This paper introduces a novel interpretable deep denoiser built by unrolling a MAP solution with a graph Laplacian prior, combining theoretical insights with neural network implementation for efficient image denoising.

Contribution

The paper presents a new graph-based deep denoiser framework that unrolls MAP solutions with a graph Laplacian regularizer, enhancing interpretability and efficiency.

Findings

Achieves competitive denoising performance with fewer parameters

Demonstrates robustness to covariate shift

Provides a graph-interpretable neural network architecture

Abstract

An image denoiser can be used for a wide range of restoration problems via the Plug-and-Play (PnP) architecture. In this paper, we propose a general framework to build an interpretable graph-based deep denoiser (GDD) by unrolling a solution to a maximum a posteriori (MAP) problem equipped with a graph Laplacian regularizer (GLR) as signal prior. Leveraging a recent theorem showing that any (pseudo-)linear denoiser $\boldsymbol \Psi$, under mild conditions, can be mapped to a solution of a MAP denoising problem regularized using GLR, we first initialize a graph Laplacian matrix $\mathbf L$ via truncated Taylor Series Expansion (TSE) of $\boldsymbol \Psi^{-1}$. Then, we compute the MAP linear system solution by unrolling iterations of the conjugate gradient (CG) algorithm into a sequence of neural layers as a feed-forward network -- one that is amenable to parameter tuning. The resulting GDD network is "graph-interpretable", low in parameter count, and easy to initialize thanks to $\mathbf L$ derived from a known well-performing denoiser $\boldsymbol \Psi$. Experimental results show that GDD achieves competitive image denoising performance compared to competitors, but employing far fewer parameters, and is more robust to covariate shift.