High Perceptual Quality Image Denoising with a Posterior Sampling CGAN

TL;DR

This paper introduces a stochastic posterior sampling CGAN-based denoiser that produces sharp, perceptually high-quality images with acceptable distortion, especially effective in high noise scenarios.

Contribution

It presents a novel CGAN framework with a theoretically grounded penalty that emphasizes perceptual quality over distortion for image denoising.

Findings

Produces vivid, diverse denoised images in high noise levels

Achieves high perceptual quality with acceptable distortion

Outperforms traditional distortion-focused methods

Abstract

The vast work in Deep Learning (DL) has led to a leap in image denoising research. Most DL solutions for this task have chosen to put their efforts on the denoiser's architecture while maximizing distortion performance. However, distortion driven solutions lead to blurry results with sub-optimal perceptual quality, especially in immoderate noise levels. In this paper we propose a different perspective, aiming to produce sharp and visually pleasing denoised images that are still faithful to their clean sources. Formally, our goal is to achieve high perceptual quality with acceptable distortion. This is attained by a stochastic denoiser that samples from the posterior distribution, trained as a generator in the framework of conditional generative adversarial networks (CGAN). Contrary to distortion-based regularization terms that conflict with perceptual quality, we introduce to the CGAN objective a theoretically founded penalty term that does not force a distortion requirement on individual samples, but rather on their mean. We showcase our proposed method with a novel denoiser architecture that achieves the reformed denoising goal and produces vivid and diverse outcomes in immoderate noise levels.