FreqNet: A Frequency-domain Image Super-Resolution Network with Dicrete Cosine Transform

TL;DR

FreqNet introduces a frequency-domain approach for image super-resolution using DCT, explicitly learning high-frequency details to improve perceptual quality and fidelity, and can enhance existing spatial SR models.

Contribution

The paper proposes FreqNet, a novel frequency-domain super-resolution network utilizing DCT and a specialized frequency loss, offering explicit high-frequency detail learning and compatibility with spatial SR models.

Findings

Explicit high-frequency detail learning improves perceptual quality.

FreqNet can be integrated with existing spatial SR models.

Frequency domain approach enhances fidelity of super-resolved images.

Abstract

Single image super-resolution(SISR) is an ill-posed problem that aims to obtain high-resolution (HR) output from low-resolution (LR) input, during which extra high-frequency information is supposed to be added to improve the perceptual quality. Existing SISR works mainly operate in the spatial domain by minimizing the mean squared reconstruction error. Despite the high peak signal-to-noise ratios(PSNR) results, it is difficult to determine whether the model correctly adds desired high-frequency details. Some residual-based structures are proposed to guide the model to focus on high-frequency features implicitly. However, how to verify the fidelity of those artificial details remains a problem since the interpretation from spatial-domain metrics is limited. In this paper, we propose FreqNet, an intuitive pipeline from the frequency domain perspective, to solve this problem. Inspired by existing frequency-domain works, we convert images into discrete cosine transform (DCT) blocks, then reform them to obtain the DCT feature maps, which serve as the input and target of our model. A specialized pipeline is designed, and we further propose a frequency loss function to fit the nature of our frequency-domain task. Our SISR method in the frequency domain can learn the high-frequency information explicitly, provide fidelity and good perceptual quality for the SR images. We further observe that our model can be merged with other spatial super-resolution models to enhance the quality of their original SR output.