Deep Iterative Residual Convolutional Network for Single Image Super-Resolution

TL;DR

This paper introduces a deep iterative residual convolutional network for single image super-resolution that leverages image regularization and iterative training to improve results with fewer parameters.

Contribution

The proposed ISRResCNet integrates iterative training and residual learning, effectively utilizing image models to enhance super-resolution performance with fewer parameters.

Findings

Outperforms state-of-the-art methods on multiple benchmarks.

Requires fewer trainable parameters for comparable or better results.

Effective across various scaling factors.

Abstract

Deep convolutional neural networks (CNNs) have recently achieved great success for single image super-resolution (SISR) task due to their powerful feature representation capabilities. The most recent deep learning based SISR methods focus on designing deeper / wider models to learn the non-linear mapping between low-resolution (LR) inputs and high-resolution (HR) outputs. These existing SR methods do not take into account the image observation (physical) model and thus require a large number of network's trainable parameters with a great volume of training data. To address these issues, we propose a deep Iterative Super-Resolution Residual Convolutional Network (ISRResCNet) that exploits the powerful image regularization and large-scale optimization techniques by training the deep network in an iterative manner with a residual learning approach. Extensive experimental results on various super-resolution benchmarks demonstrate that our method with a few trainable parameters improves the results for different scaling factors in comparison with the state-of-art methods.