LAPAR: Linearly-Assembled Pixel-Adaptive Regression Network for Single Image Super-Resolution and Beyond

TL;DR

LAPAR introduces a lightweight, pixel-adaptive regression network for single image super-resolution that models LR to HR mapping as a linear coefficient regression, achieving state-of-the-art results efficiently and extending to other restoration tasks.

Contribution

The paper proposes a novel linear coefficient regression approach over predefined filter bases, making super-resolution models more lightweight and versatile for various image restoration tasks.

Findings

Achieves state-of-the-art super-resolution performance.

Extends effectively to denoising and JPEG deblocking.

Model is highly lightweight and easy to optimize.

Abstract

Single image super-resolution (SISR) deals with a fundamental problem of upsampling a low-resolution (LR) image to its high-resolution (HR) version. Last few years have witnessed impressive progress propelled by deep learning methods. However, one critical challenge faced by existing methods is to strike a sweet spot of deep model complexity and resulting SISR quality. This paper addresses this pain point by proposing a linearly-assembled pixel-adaptive regression network (LAPAR), which casts the direct LR to HR mapping learning into a linear coefficient regression task over a dictionary of multiple predefined filter bases. Such a parametric representation renders our model highly lightweight and easy to optimize while achieving state-of-the-art results on SISR benchmarks. Moreover, based on the same idea, LAPAR is extended to tackle other restoration tasks, e.g., image denoising and JPEG image deblocking, and again, yields strong performance. The code is available at https://github.com/dvlab-research/Simple-SR.