Wide Inference Network for Image Denoising via Learning Pixel-distribution Prior

TL;DR

This paper introduces Wide Inference Networks (WIN), a CNN-based approach that leverages wider networks to learn pixel-distribution priors for improved image denoising, outperforming existing methods.

Contribution

The paper proposes a novel strategy of increasing CNN width to learn pixel-distribution features, revealing that wider CNNs better capture noise priors for denoising tasks.

Findings

Wider CNNs learn more accurate pixel-distribution features.

WIN outperforms state-of-the-art denoising methods on AWGN.

Learning noise priors is effective for image denoising.

Abstract

We explore an innovative strategy for image denoising by using convolutional neural networks (CNN) to learn similar pixel-distribution features from noisy images. Many types of image noise follow a certain pixel-distribution in common, such as additive white Gaussian noise (AWGN). By increasing CNN's width with larger reception fields and more channels in each layer, CNNs can reveal the ability to extract more accurate pixel-distribution features. The key to our approach is a discovery that wider CNNs with more convolutions tend to learn the similar pixel-distribution features, which reveals a new strategy to solve low-level vision problems effectively that the inference mapping primarily relies on the priors behind the noise property instead of deeper CNNs with more stacked nonlinear layers. We evaluate our work, Wide inference Networks (WIN), on AWGN and demonstrate that by learning pixel-distribution features from images, WIN-based network consistently achieves significantly better performance than current state-of-the-art deep CNN-based methods in both quantitative and visual evaluations. \textit{Code and models are available at \url{https://github.com/cswin/WIN}}.