Although recent diffusion-based single-step super-resolution methods achieve better performance as compared to SinSR, they are computationally complex. To improve the performance of SinSR, we investigate preserving the high-frequency detail features during super-resolution (SR) because the downgraded images lack detailed information. For this purpose, we introduce a high-frequency perceptual loss by utilizing an invertible neural network (INN) pretrained on the ImageNet dataset. Different feature maps of pretrained INN produce different high-frequency aspects of an image. During the training phase, we impose to preserve the high-frequency features of super-resolved and ground truth (GT) images that improve the SR image quality during inference. Furthermore, we also utilize the Jenson-Shannon divergence between GT and SR images in the pretrained DINO-v2 embedding space to match their distribution. By introducing the - preserving loss and distribution matching constraint in the single-step SR (), we achieve a state-of-the-art CLIPIQA score in the benchmark RealSR, RealSet65, DIV2K-Val, and ImageNet datasets. Furthermore, the experimental results in several datasets demonstrate that our high-frequency perceptual loss yields better SR image quality than LPIPS and VGG-based perceptual losses. Our code will be released at https://github.com/shoaib-sami/HF-Diff.
View on arXiv@article{sami2025_2411.13548,
title={ MGHF: Multi-Granular High-Frequency Perceptual Loss for Image Super-Resolution },
author={ Shoaib Meraj Sami and Md Mahedi Hasan and Mohammad Saeed Ebrahimi Saadabadi and Jeremy Dawson and Nasser Nasrabadi and Raghuveer Rao },
journal={arXiv preprint arXiv:2411.13548},
year={ 2025 }
}