Fully Trainable and Interpretable Non-Local Sparse Models for Image Restoration
This provides a compact and interpretable solution for image restoration tasks, addressing the need for efficient models in computer vision.
The paper tackles image restoration by proposing a trainable and interpretable non-local sparse model, achieving performance on par or better than state-of-the-art methods with only 100K parameters in tasks like denoising, jpeg deblocking, and demosaicking.
Non-local self-similarity and sparsity principles have proven to be powerful priors for natural image modeling. We propose a novel differentiable relaxation of joint sparsity that exploits both principles and leads to a general framework for image restoration which is (1) trainable end to end, (2) fully interpretable, and (3) much more compact than competing deep learning architectures. We apply this approach to denoising, jpeg deblocking, and demosaicking, and show that, with as few as 100K parameters, its performance on several standard benchmarks is on par or better than state-of-the-art methods that may have an order of magnitude or more parameters.