NTIRE 2020 真实图像去噪挑战赛：数据集、方法与结果
计算机视觉与模式识别
2020-05-11 v1 图像与视频处理
摘要
本文回顾了 NTIRE 2020 真实图像去噪挑战赛，重点关注新引入的数据集、所提出的方法及其结果。该挑战赛是之前基于 SIDD 基准的 NTIRE 2019 真实图像去噪挑战赛的新版本。本次挑战赛基于新收集的验证和测试图像数据集，因此命名为 SIDD+。该挑战赛设有两个赛道，用于定量评估在 (1) Bayer 模式 rawRGB 和 (2) 标准 RGB (sRGB) 色彩空间下的图像去噪性能。每个赛道约有 250 名注册参与者。共有 22 支队伍提出了 24 种方法，在挑战赛的最终阶段展开角逐。参赛队伍所提出的方法代表了针对真实噪声图像的图像去噪领域的当前 SOTA 性能。新收集的 SIDD+ 数据集公开获取地址为：https://bit.ly/siddplus_data。
引用
@article{arxiv.2005.04117,
title = {NTIRE 2020 Challenge on Real Image Denoising: Dataset, Methods and Results},
author = {Abdelrahman Abdelhamed and Mahmoud Afifi and Radu Timofte and Michael S. Brown and Yue Cao and Zhilu Zhang and Wangmeng Zuo and Xiaoling Zhang and Jiye Liu and Wendong Chen and Changyuan Wen and Meng Liu and Shuailin Lv and Yunchao Zhang and Zhihong Pan and Baopu Li and Teng Xi and Yanwen Fan and Xiyu Yu and Gang Zhang and Jingtuo Liu and Junyu Han and Errui Ding and Songhyun Yu and Bumjun Park and Jechang Jeong and Shuai Liu and Ziyao Zong and Nan Nan and Chenghua Li and Zengli Yang and Long Bao and Shuangquan Wang and Dongwoon Bai and Jungwon Lee and Youngjung Kim and Kyeongha Rho and Changyeop Shin and Sungho Kim and Pengliang Tang and Yiyun Zhao and Yuqian Zhou and Yuchen Fan and Thomas Huang and Zhihao Li and Nisarg A. Shah and Wei Liu and Qiong Yan and Yuzhi Zhao and Marcin Możejko and Tomasz Latkowski and Lukasz Treszczotko and Michał Szafraniuk and Krzysztof Trojanowski and Yanhong Wu and Pablo Navarrete Michelini and Fengshuo Hu and Yunhua Lu and Sujin Kim and Wonjin Kim and Jaayeon Lee and Jang-Hwan Choi and Magauiya Zhussip and Azamat Khassenov and Jong Hyun Kim and Hwechul Cho and Priya Kansal and Sabari Nathan and Zhangyu Ye and Xiwen Lu and Yaqi Wu and Jiangxin Yang and Yanlong Cao and Siliang Tang and Yanpeng Cao and Matteo Maggioni and Ioannis Marras and Thomas Tanay and Gregory Slabaugh and Youliang Yan and Myungjoo Kang and Han-Soo Choi and Kyungmin Song and Shusong Xu and Xiaomu Lu and Tingniao Wang and Chunxia Lei and Bin Liu and Rajat Gupta and Vineet Kumar},
journal= {arXiv preprint arXiv:2005.04117},
year = {2020}
}