A Sub-band Approach to Deep Denoising Wavelet Networks and a Frequency-adaptive Loss for Perceptual Quality

TL;DR

This paper introduces a novel sub-band approach to deep wavelet-based denoising networks and a frequency-adaptive loss function that enhances perceptual quality by focusing on the most error-prone frequencies.

Contribution

It presents a new method of applying separate convolutional layers to each wavelet sub-band and a frequency-based loss that improves denoising accuracy and perceptual quality.

Findings

Improved denoising accuracy with sub-band convolution approach.

Enhanced perceptual quality using frequency-adaptive loss.

Better focus on problematic frequency components during training.

Abstract

In this paper, we propose two contributions to neural network based denoising. First, we propose applying separate convolutional layers to each sub-band of discrete wavelet transform (DWT) as opposed to the common usage of DWT which concatenates all sub-bands and applies a single convolution layer. We show that our approach to using DWT in neural networks improves the accuracy notably, due to keeping the sub-band order uncorrupted prior to inverse DWT. Our second contribution is a denoising loss based on top k-percent of errors in frequency domain. A neural network trained with this loss, adaptively focuses on frequencies that it fails to recover the most in each iteration. We show that this loss results into better perceptual quality by providing an image that is more balanced in terms of the errors in frequency components.