LINN: Lifting Inspired Invertible Neural Network for Image Denoising

TL;DR

This paper introduces LINN, an invertible neural network inspired by wavelet lifting schemes, designed for image denoising, which achieves comparable results to existing methods with fewer parameters.

Contribution

The paper presents a novel invertible neural network architecture for image denoising that leverages wavelet-inspired lifting schemes and sparsity-driven denoising, reducing parameter count.

Findings

Achieves denoising performance comparable to DnCNN.

Uses only one-quarter of the learnable parameters.

Employs a wavelet-inspired invertible architecture.

Abstract

In this paper, we propose an invertible neural network for image denoising (DnINN) inspired by the transform-based denoising framework. The proposed DnINN consists of an invertible neural network called LINN whose architecture is inspired by the lifting scheme in wavelet theory and a sparsity-driven denoising network which is used to remove noise from the transform coefficients. The denoising operation is performed with a single soft-thresholding operation or with a learned iterative shrinkage thresholding network. The forward pass of LINN produces an over-complete representation which is more suitable for denoising. The denoised image is reconstructed using the backward pass of LINN using the output of the denoising network. The simulation results show that the proposed DnINN method achieves results comparable to the DnCNN method while only requiring 1/4 of learnable parameters.