Class-Aware Fully-Convolutional Gaussian and Poisson Denoising
This work addresses image denoising for applications requiring high-quality visual outputs, but it appears incremental as it builds on existing neural network approaches with added class-awareness.
The authors tackled image denoising by proposing a fully-convolutional neural network that exploits the gradual nature of denoising and incorporates semantic class information, advancing state-of-the-art performance for Gaussian and Poisson noise.
We propose a fully-convolutional neural-network architecture for image denoising which is simple yet powerful. Its structure allows to exploit the gradual nature of the denoising process, in which shallow layers handle local noise statistics, while deeper layers recover edges and enhance textures. Our method advances the state-of-the-art when trained for different noise levels and distributions (both Gaussian and Poisson). In addition, we show that making the denoiser class-aware by exploiting semantic class information boosts performance, enhances textures and reduces artifacts.