MMSR: Multiple-Model Learned Image Super-Resolution Benefiting From Class-Specific Image Priors

TL;DR

This paper introduces MMSR, a multi-model super-resolution framework that leverages class-specific image priors and a fusion network to outperform single-model SR approaches in diverse image content scenarios.

Contribution

The work proposes training multiple SR models for different image classes and a fusion network, significantly improving super-resolution performance over existing single-model methods.

Findings

MMSR outperforms state-of-the-art single-model SR methods quantitatively.

MMSR provides better visual quality in super-resolved images.

The fusion network effectively combines outputs from multiple models.

Abstract

Assuming a known degradation model, the performance of a learned image super-resolution (SR) model depends on how well the variety of image characteristics within the training set matches those in the test set. As a result, the performance of an SR model varies noticeably from image to image over a test set depending on whether characteristics of specific images are similar to those in the training set or not. Hence, in general, a single SR model cannot generalize well enough for all types of image content. In this work, we show that training multiple SR models for different classes of images (e.g., for text, texture, etc.) to exploit class-specific image priors and employing a post-processing network that learns how to best fuse the outputs produced by these multiple SR models surpasses the performance of state-of-the-art generic SR models. Experimental results clearly demonstrate that the proposed multiple-model SR (MMSR) approach significantly outperforms a single pre-trained state-of-the-art SR model both quantitatively and visually. It even exceeds the performance of the best single class-specific SR model trained on similar text or texture images.