Unsupervised Image Super-Resolution with an Indirect Supervised Path
This addresses the challenge of applying super-resolution to real-world images without paired training data, offering a flexible solution for balancing distortion and perceptual quality, though it is incremental as it builds on existing unsupervised translation and supervised super-resolution methods.
The paper tackles the problem of single image super-resolution on real data by proposing a two-stage framework that uses unsupervised image translation to bridge real low-resolution images to synthetic ones, enabling indirect supervised super-resolution, and achieves favorable performance on NTIRE 2017 and 2018 datasets compared to supervised methods.
The task of single image super-resolution (SISR) aims at reconstructing a high-resolution (HR) image from a low-resolution (LR) image. Although significant progress has been made by deep learning models, they are trained on synthetic paired data in a supervised way and do not perform well on real data. There are several attempts that directly apply unsupervised image translation models to address such a problem. However, unsupervised low-level vision problem poses more challenge on the accuracy of translation. In this work,we propose a novel framework which is composed of two stages: 1) unsupervised image translation between real LR images and synthetic LR images; 2) supervised super-resolution from approximated real LR images to HR images. It takes the synthetic LR images as a bridge and creates an indirect supervised path from real LR images to HR images. Any existed deep learning based image super-resolution model can be integrated into the second stage of the proposed framework for further improvement. In addition it shows great flexibility in balancing between distortion and perceptual quality under unsupervised setting. The proposed method is evaluated on both NTIRE 2017 and 2018 challenge datasets and achieves favorable performance against supervised methods.