학술논문
NTIRE 2022 Spectral Demosaicing Challenge and Data Set
Document Type
Conference
Author
Arad, Boaz; Timofte, Radu; Yahel, Rony; Morag, Nimrod; Bernat, Amir; Wu, Yaqi; Wu, Xun; Fan, Zhihao; Xia, Chenjie; Zhang, Feng; Liu, Shuai; Li, Yongqiang; Feng, Chaoyu; Lei, Lei; Zhang, Mingwei; Feng, Kai; Zhang, Xun; Yao, Jiaxin; Zhao, Yongqiang; Ma, Suina; He, Fan; Dong, Yangyang; Yu, Shufang; Qiu, Difa; Liu, Jinhui; Bi, Mengzhao; Song, Beibei; Sun, WenFang; Zheng, Jiesi; Zhao, Bowen; Cao, Yanpeng; Yang, Jiangxin; Cao, Yanlong; Kong, Xiangyu; Yu, Jingbo; Xue, Yuanyang; Xie, Zheng
Source
2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) CVPRW Computer Vision and Pattern Recognition Workshops (CVPRW), 2022 IEEE/CVF Conference on. :881-895 Jun, 2022
Subject
Language
ISSN
2160-7516
Abstract
This paper presents the first challenge on demosaicing of natural spectral images for snapshot hyperspectral imaging systems (HIS) which utilize a multi-spectral filer array (MSFA), i.e., the recovery of whole-scene hyperspectral information from spatially sub-sampled hyperspectral information. This challenge expands the "ARAD_1K" data set to a first-of-its-kind large-scale data set for multispectral filter array demosaicing of natural scenes containing 1,000 images. Challenge participants were required to recover hyperspectral information from synthetically generated MSFA images simulating capture by a known calibrated snapshot mosaic hyperspectral camera. The challenge was attended by 157 teams, with 29 teams competing in the final testing phase, 7 of which provided detailed descriptions of their methodology which are included in this report. The performance of these submissions is reviewed and provided here as a gauge for the current state-of-the-art in multi-spectral filter array demosaicing of natural images.