FS-NCSR: Increasing Diversity of the Super-Resolution Space via Frequency Separation and Noise-Conditioned Normalizing Flow

TL;DR

FS-NCSR enhances super-resolution diversity and quality by leveraging frequency separation and noise conditioning, effectively estimating high-frequency details without redundant low-frequency info, outperforming previous flow-based models.

Contribution

The paper introduces FS-NCSR, a novel flow-based super-resolution method that improves diversity and quality by focusing on high-frequency information through frequency separation and noise conditioning.

Findings

Significantly improves diversity score over previous models.

Maintains high image quality with reduced artifacts.

Outperforms NTIRE 2021 challenge winner NCSR.

Abstract

Super-resolution suffers from an innate ill-posed problem that a single low-resolution (LR) image can be from multiple high-resolution (HR) images. Recent studies on the flow-based algorithm solve this ill-posedness by learning the super-resolution space and predicting diverse HR outputs. Unfortunately, the diversity of the super-resolution outputs is still unsatisfactory, and the outputs from the flow-based model usually suffer from undesired artifacts which causes low-quality outputs. In this paper, we propose FS-NCSR which produces diverse and high-quality super-resolution outputs using frequency separation and noise conditioning compared to the existing flow-based approaches. As the sharpness and high-quality detail of the image rely on its high-frequency information, FS-NCSR only estimates the high-frequency information of the high-resolution outputs without redundant low-frequency components. Through this, FS-NCSR significantly improves the diversity score without significant image quality degradation compared to the NCSR, the winner of the previous NTIRE 2021 challenge.