Fast, nonlocal and neural: a lightweight high quality solution to image denoising
This addresses the problem of efficient, high-quality image denoising for applications such as mobile terminals, though it is incremental as it builds on existing nonlocal and CNN methods.
The paper tackles the computational cost and over-smoothing issues of CNNs in image denoising by combining a nonlocal algorithm with a lightweight residual CNN, achieving a 10-20 times speedup over CNNs with equivalent performance and higher PSNR, especially on complex textures like in the MIT Moire dataset.
With the widespread application of convolutional neural networks (CNNs), the traditional model based denoising algorithms are now outperformed. However, CNNs face two problems. First, they are computationally demanding, which makes their deployment especially difficult for mobile terminals. Second, experimental evidence shows that CNNs often over-smooth regular textures present in images, in contrast to traditional non-local models. In this letter, we propose a solution to both issues by combining a nonlocal algorithm with a lightweight residual CNN. This solution gives full latitude to the advantages of both models. We apply this framework to two GPU implementations of classic nonlocal algorithms (NLM and BM3D) and observe a substantial gain in both cases, performing better than the state-of-the-art with low computational requirements. Our solution is between 10 and 20 times faster than CNNs with equivalent performance and attains higher PSNR. In addition the final method shows a notable gain on images containing complex textures like the ones of the MIT Moire dataset.