SRTGAN: Triplet Loss Based Generative Adversarial Network for Real-World Super-Resolution

EasyChair Preprint 7807, version 2

Abstract

Many applications such as surveillance, forensics, satellite imaging, medical imaging, etc., demand High-Resolution (HR) images. However, obtaining an HR image is not always possible due to the limitations of optical sensors and their costs. An alternative solution called Single Image Super-Resolution (SISR) is a software-driven approach that aims to take a Low-Resolution (LR) image and obtain the HR image. Most supervised SISR solutions use the HR image as a target and do not include the information provided in the LR image, which could be valuable. In this work, we introduce Triplet Loss-based Generative Adversarial Network hereafter referred to as SRTGAN for image SR problem on real-world degradation. We introduce a new triplet-based adversarial loss function which exploits the information provided in the LR image by using it as a negative sample. Allowing the patch-based discriminator with access to both HR and LR images optimizes to better differentiate between HR and LR images; hence, improving the adversary. Further, we propose to fuse the adversarial loss, content loss, perceptual loss, and quality loss to obtain an SR image with high perceptual fidelity. We validate the superior performance of the proposed method over the other existing methods on the RealSR dataset in terms of quantitative and qualitative metrics.

Keyphrases: Generative Adversarial Networks, deep learning, image super-resolution

@booklet{EasyChair:7807,
  author    = {Dhruv Patel and Abhinav Jain and Simran Bawkar and Manav Khorasiya and Kalpesh Prajapati and Kishor Upla and Kiran Raja and R. Raghavendra and Christoph Busch},
  title     = {SRTGAN: Triplet Loss Based Generative Adversarial Network for Real-World Super-Resolution},
  howpublished = {EasyChair Preprint 7807},
  year      = {EasyChair, 2022}}