Controllable Confidence-Based Image Denoising

TL;DR

This paper introduces a controllable image denoising framework that combines generic and deep learning methods, allowing users to adjust confidence levels and improve interpretability while enhancing generalization and robustness.

Contribution

The novel framework fuses generic and deep learning denoisers in the frequency domain and estimates confidence to improve interpretability and robustness against out-of-distribution inputs.

Findings

Effective noise removal across various scenarios

Enhanced interpretability through confidence estimation

Robust performance on out-of-distribution images

Abstract

Image denoising is a classic restoration problem. Yet, current deep learning methods are subject to the problems of generalization and interpretability. To mitigate these problems, in this project, we present a framework that is capable of controllable, confidence-based noise removal. The framework is based on the fusion between two different denoised images, both derived from the same noisy input. One of the two is denoised using generic algorithms (e.g. Gaussian), which make few assumptions on the input images, therefore, generalize in all scenarios. The other is denoised using deep learning, performing well on seen datasets. We introduce a set of techniques to fuse the two components smoothly in the frequency domain. Beyond that, we estimate the confidence of a deep learning denoiser to allow users to interpret the output, and provide a fusion strategy that safeguards them against out-of-distribution inputs. Through experiments, we demonstrate the effectiveness of the proposed framework in different use cases.