Asymmetric CNN for image super-resolution

TL;DR

This paper introduces ACNet, an asymmetric CNN architecture designed for image super-resolution, which emphasizes local salient features and enhances feature fusion to improve super-resolution quality and efficiency.

Contribution

The paper proposes a novel asymmetric CNN with specialized blocks for better feature extraction and fusion in image super-resolution tasks, addressing local feature importance and long-term dependencies.

Findings

ACNet outperforms existing methods in super-resolution accuracy.

The asymmetric and memory enhancement blocks improve feature representation.

ACNet effectively handles blind super-resolution and noise scenarios.

Abstract

Deep convolutional neural networks (CNNs) have been widely applied for low-level vision over the past five years. According to nature of different applications, designing appropriate CNN architectures is developed. However, customized architectures gather different features via treating all pixel points as equal to improve the performance of given application, which ignores the effects of local power pixel points and results in low training efficiency. In this paper, we propose an asymmetric CNN (ACNet) comprising an asymmetric block (AB), a memory enhancement block (MEB) and a high-frequency feature enhancement block (HFFEB) for image super-resolution. The AB utilizes one-dimensional asymmetric convolutions to intensify the square convolution kernels in horizontal and vertical directions for promoting the influences of local salient features for SISR. The MEB fuses all hierarchical low-frequency features from the AB via residual learning (RL) technique to resolve the long-term dependency problem and transforms obtained low-frequency features into high-frequency features. The HFFEB exploits low- and high-frequency features to obtain more robust super-resolution features and address excessive feature enhancement problem. Addditionally, it also takes charge of reconstructing a high-resolution (HR) image. Extensive experiments show that our ACNet can effectively address single image super-resolution (SISR), blind SISR and blind SISR of blind noise problems. The code of the ACNet is shown at https://github.com/hellloxiaotian/ACNet.