Alternative design of DeepPDNet in the context of image restoration

TL;DR

This paper introduces a novel deep neural network architecture for image restoration based on unfolded Chambolle-Pock primal-dual iterations, optimizing parameters like step-sizes and operators for improved performance.

Contribution

It presents a new deep network design that unrolls Chambolle-Pock iterations, integrating primal-dual optimization into a learnable deep architecture for image restoration.

Findings

Demonstrates effective image restoration on BSD68 database.

Shows good convergence behavior of the proposed deep primal-dual network.

Provides a fully detailed backpropagation procedure for training.

Abstract

This work designs an image restoration deep network relying on unfolded Chambolle-Pock primal-dual iterations. Each layer of our network is built from Chambolle-Pock iterations when specified for minimizing a sum of a $\ell_2$-norm data-term and an analysis sparse prior. The parameters of our network are the step-sizes of the Chambolle-Pock scheme and the linear operator involved in sparsity-based penalization, including implicitly the regularization parameter. A backpropagation procedure is fully described. Preliminary experiments illustrate the good behavior of such a deep primal-dual network in the context of image restoration on BSD68 database.