Image Super-Resolution Using Very Deep Residual Channel Attention Networks

TL;DR

This paper introduces RCAN, a very deep residual network with channel attention for image super-resolution, effectively focusing on high-frequency details and outperforming existing methods in accuracy and visual quality.

Contribution

The paper presents a novel very deep residual network with channel attention and residual in residual structure for improved image super-resolution.

Findings

RCAN achieves superior accuracy over state-of-the-art methods.

RCAN produces better visual quality in super-resolved images.

The channel attention mechanism effectively enhances feature representation.

Abstract

Convolutional neural network (CNN) depth is of crucial importance for image super-resolution (SR). However, we observe that deeper networks for image SR are more difficult to train. The low-resolution inputs and features contain abundant low-frequency information, which is treated equally across channels, hence hindering the representational ability of CNNs. To solve these problems, we propose the very deep residual channel attention networks (RCAN). Specifically, we propose a residual in residual (RIR) structure to form very deep network, which consists of several residual groups with long skip connections. Each residual group contains some residual blocks with short skip connections. Meanwhile, RIR allows abundant low-frequency information to be bypassed through multiple skip connections, making the main network focus on learning high-frequency information. Furthermore, we propose a channel attention mechanism to adaptively rescale channel-wise features by considering interdependencies among channels. Extensive experiments show that our RCAN achieves better accuracy and visual improvements against state-of-the-art methods.