Training Deep Learning Based Denoisers without Ground Truth Data
This addresses a practical limitation in denoising applications where obtaining clean data is challenging, offering a solution for scenarios like medical imaging or astronomy, though it is incremental as it builds on existing SURE techniques.
The paper tackles the problem of training deep learning denoisers without requiring noiseless ground truth data, which is often unavailable, by proposing a method based on Stein's unbiased risk estimator (SURE) that uses only noisy images with Gaussian noise, achieving performances close to networks trained with ground truth and outperforming conventional methods like BM3D and deep image prior in some cases.
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.