Deep Iterative Residual Convolutional Network for Single Image Super-Resolution
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.
Cite
@article{arxiv.2009.04809,
title = {Deep Iterative Residual Convolutional Network for Single Image Super-Resolution},
author = {Rao Muhammad Umer and Gian Luca Foresti and Christian Micheloni},
journal= {arXiv preprint arXiv:2009.04809},
year = {2020}
}
Comments
To be appeared in proceedings of the 25th IEEE International Conference on Pattern Recognition (ICPR). arXiv admin note: text overlap with arXiv:2005.00953, arXiv:2009.03693