Training Deep Learning Based Denoisers without Ground Truth Data

TL;DR

This paper introduces a SURE-based training method for deep learning denoisers that does not require ground truth images, enabling effective training solely on noisy data and achieving performance comparable to traditional methods.

Contribution

The paper presents a novel SURE-based approach for training deep neural network denoisers using only noisy images, eliminating the need for noiseless ground truth data.

Findings

SURE-based training achieves near ground-truth performance.

The method outperforms BM3D and deep image prior.

Effective for both grayscale and color images.

Abstract

Recently developed deep-learning-based denoisers often outperform state-of-the-art conventional denoisers such as the BM3D. They are typically trained to minimize the mean squared error (MSE) between the output image of a deep neural network (DNN) and a ground truth image. Thus, it is important for deep-learning-based denoisers to use high quality noiseless ground truth data for high performance. However, it is often challenging or even infeasible to obtain noiseless images in some applications. Here, we propose a method based on Stein's unbiased risk estimator (SURE) for training DNN denoisers based only on the use of noisy images in the training data with Gaussian noise. We demonstrate that our SURE-based method, without the use of ground truth data, is able to train DNN denoisers to yield performances close to those networks trained with ground truth for both grayscale and color images. We also propose a SURE-based refining method with a noisy test image for further performance improvement. Our quick refining method outperformed conventional BM3D, deep image prior, and often the networks trained with ground truth. Potential extension of our SURE-based methods to Poisson noise model was also investigated.