Image Super-Resolution via Dual-Dictionary Learning And Sparse Representation

TL;DR

This paper introduces a dual-dictionary learning approach for image super-resolution that separately reconstructs main and residual high-frequency details, significantly improving detail recovery and visual quality over existing methods.

Contribution

It proposes a novel dual-dictionary learning framework that models high-frequency details as two components, enhancing super-resolution performance beyond prior algorithms.

Findings

Better PSNR scores than state-of-the-art methods

Improved visual detail recovery in super-resolved images

Effective two-layer progressive scheme enhances results

Abstract

Learning-based image super-resolution aims to reconstruct high-frequency (HF) details from the prior model trained by a set of high- and low-resolution image patches. In this paper, HF to be estimated is considered as a combination of two components: main high-frequency (MHF) and residual high-frequency (RHF), and we propose a novel image super-resolution method via dual-dictionary learning and sparse representation, which consists of the main dictionary learning and the residual dictionary learning, to recover MHF and RHF respectively. Extensive experimental results on test images validate that by employing the proposed two-layer progressive scheme, more image details can be recovered and much better results can be achieved than the state-of-the-art algorithms in terms of both PSNR and visual perception.