NTIRE 2026 远程感知红外图像超分辨率挑战第一报：基准结果与方法概述
计算机视觉与模式识别
2026-04-24 v1 人工智能
摘要
本文介绍了 NTIRE 2026 远程感知红外图像超分辨率 (x4) 挑战，这是 NTIRE 2026 系列挑战中的一个关联项目。该挑战旨在从通过 x4 缩放因子进行的双立方 downsampling 生成的低分辨率 (LR) 输入中恢复高分辨率 (HR) 红外图像。目标是开发有效模型或解决方案，为红外图像超分辨率在远程感知场景中实现最先进的性能。为反映红外数据特征和实际应用需求，挑战采用单轨设置。共有 115 名参赛者注册，13 支团队提交了有效条目。本报告总结了挑战设计、数据集、评估协议、主要结果以及每支团队的代表性方法。该挑战作为推动红外图像超分辨率研究并促进开发适用于实际远程感知应用的有效解决方案的基准，发挥了重要作用。
引用
@article{arxiv.2604.21312,
title = {The First Challenge on Remote Sensing Infrared Image Super-Resolution at NTIRE 2026: Benchmark Results and Method Overview},
author = {Kai Liu and Haoyang Yue and Zeli Lin and Zheng Chen and Jingkai Wang and Jue Gong and Jiatong Li and Xianglong Yan and Libo Zhu and Jianze Li and Ziqing Zhang and Zihan Zhou and Xiaoyang Liu and Radu Timofte and Yulun Zhang and Junye Chen and Zhenming Yan and Yucong Hong and Ruize Han and Song Wang and Li Pang and Heng Zhao and Xinqiao Wu and Deyu Meng and Xiangyong Cao and Weijun Yuan and Zhan Li and Zhanglu Chen and Boyang Yao and Yihang Chen and Yifan Deng and Zengyuan Zuo and Junjun Jiang and Saiprasad Meesiyawar and Sulocha Yatageri and Nikhil Akalwadi and Ramesh Ashok Tabib and Uma Mudenagudi and Jiachen Tu and Yaokun Shi and Guoyi Xu and Yaoxin Jiang and Cici Liu and Tongyao Mu and Qiong Cao and Yifan Wang and Kosuke Shigematsu and Hiroto Shirono and Asuka Shin and Wei Zhou and Linfeng Li and Lingdong Kong and Ce Wang and Xingwei Zhong and Wanjie Sun and Dafeng Zhang and Hongxin Lan and Qisheng Xu and Mingyue He and Hui Geng and Tianjiao Wan and Kele Xu and Changjian Wang and Antoine Carreaud and Nicola Santacroce and Shanci Li and Jan Skaloud and Adrien Gressin},
journal= {arXiv preprint arXiv:2604.21312},
year = {2026}
}
备注
Github Repo: https://github.com/Kai-Liu001/NTIRE2026_infraredSR