Sen2CHRIS dataset for training and test
收藏资源简介：
Compact High Resolution Imaging Spectrometer, namely CHRIS, is carried on PRoject for On Board Autonomy 1 (PROBA-1) satellite and provides hyperspectral images.However, CHRIS images are always accompanied by hybrid noise, stripe gap, and high cloud cover. Furthermore, all satellite passes are systematically acquired according to a fixed acquisition plan. Observation over a new specific area should be performed by submitting the request to add a new site to the acquisition plan, which costs much money and time. To verify the model performance of solving generalized spectral super-resolution, Sen2CHRIS data set is generated from three freely available data subsets, including Xiong'an, Washington DC Mall, and Chikusei, by downsampling the spectral channels of free hyperspectral data to the same of CHRIS and Sentinel-2 using Hysure. The file named "Chikusei_CHRIS.h5" contains 1016 training samples where the high-resolution multispectral images are with size of 128×128×4 while the size of low-resolution multispectral images is 64×64×4. Moreover, we also give some test images in the "Chikusei_test.mat". The file named "DCMall_CHRIS.h5" contains 2424 training samples where the high-resolution multispectral images are with size of 64×64×4 while the size of low-resolution multispectral images is 32×32×4. Moreover, we also give some test images in the "DCMall_test.mat". The file named "Xiongan_CHRIS.h5" contains 1216 training samples where the high-resolution multispectral images are with size of 128×128×4 while the size of low-resolution multispectral images is 64×64×4. Moreover, we also give some test images in the "Xiongan_test.mat". Details about generalized spectral super-resolution can be found in the following paper. Paper: J. He, Q. Yuan, J. Li, and L. Zhang, "PoNet: A universal physical optimization-based spectral super-resolution network for arbitrary multispectral images," Information Fusion, vol. 80, pp. 205–225, 2022. More information about the author can be found at https://jianghe96.github.io/ If this dataset is helpful please cite as: @article{He2022PoNet, title={PoNet: A universal physical optimization-based spectral super-resolution network for arbitrary multispectral images}, author={He, Jiang and Yuan, Qiangqiang and Li, Jie and Zhang, Liangpei}, journal={Information Fusion}, volume={80}, pages={205--225}, year={2022}, }
紧凑型高分辨率成像光谱仪（Compact High Resolution Imaging Spectrometer, CHRIS）搭载于在轨自主1号（PROBA-1）卫星，可获取高光谱图像。然而，CHRIS图像常伴随混合噪声、条带间隙与高云量干扰。此外，所有卫星过境采集均按照固定采集计划系统执行，若需对新的特定区域进行观测，需提交申请将该区域新增至采集计划中，这将耗费大量资金与时间。为验证模型在广义光谱超分辨率任务中的性能，研究人员通过Hysure工具将公开高光谱数据的光谱通道降采样至与CHRIS和Sentinel-2一致的波段规格，从三个公开可用的数据子集——雄安（Xiong'an）、华盛顿特区购物中心（Washington DC Mall）与筑波（Chikusei）——中生成了Sen2CHRIS数据集。 名为"Chikusei_CHRIS.h5"的文件包含1016个训练样本，其中高分辨率多光谱图像尺寸为128×128×4，低分辨率多光谱图像尺寸为64×64×4；此外，"Chikusei_test.mat"中提供了部分测试图像。 名为"DCMall_CHRIS.h5"的文件包含2424个训练样本，其中高分辨率多光谱图像尺寸为64×64×4，低分辨率多光谱图像尺寸为32×32×4；此外，"DCMall_test.mat"中提供了部分测试图像。 名为"Xiongan_CHRIS.h5"的文件包含1216个训练样本，其中高分辨率多光谱图像尺寸为128×128×4，低分辨率多光谱图像尺寸为64×64×4；此外，"Xiongan_test.mat"中提供了部分测试图像。 关于广义光谱超分辨率的详细内容可参阅以下论文： 论文：J. He, Q. Yuan, J. Li 与 L. Zhang, "PoNet: 一种基于物理优化的通用光谱超分辨率网络，适用于任意多光谱图像", Information Fusion, 第80卷, 第205–225页, 2022年。 更多作者相关信息可访问：https://jianghe96.github.io/ 若该数据集对你有所帮助，请引用如下文献： @article{He2022PoNet, title={PoNet: A universal physical optimization-based spectral super-resolution network for arbitrary multispectral images}, author={He, Jiang and Yuan, Qiangqiang and Li, Jie and Zhang, Liangpei}, journal={Information Fusion}, volume={80}, pages={205--225}, year={2022}, }
