CRA-MHIF:Consensus Reasoning Attention Network for Multispectral and Hyperspectral Image Fusion

Experimental Data for CRA-MHIF: Consensus Reasoning Attention Network for Multispectral and Hyperspectral Image FusionThis package includes all supporting data for the paper:Consensus Reasoning Attention Network for Multispectral and Hyperspectral Image Fusion1. Datasets UsedAll datasets are publicly available and used as described in the paper.CAVE Dataset:Source: https://cave.cs.columbia.edu/repository/MultispectralDescription: 32 indoor multispectral scenes with 31 spectral bands (400–700 nm, 10 nm step), spatial resolution 512×512.Pavia University Dataset (PaviaU):Source: https://www.ehu.eus/ccwintco/index.php/Hyperspectral_Remote_Sensing_Scenes#Pavia_Centre_and_UniversityDescription: Hyperspectral image acquired by ROSIS-03 sensor over Pavia, Italy. Contains 103 valid spectral bands (0.43–0.86 μm), spatial resolution 1.3 m.Washington DC Mall Dataset (WDCM):Source: https://engineering.purdue.edu/~biehl/MultiSpec/hyperspectral.htmlDescription: Hyperspectral image captured by HYDICE airborne sensor. Contains 191 valid spectral bands after removing water vapor absorption bands, spatial resolution 2.5 m.Yellow River Estuary Dataset (YRE):Source:YRE-Dataset/Description: Real-world remote sensing dataset composed of Gaofen-5 AHSI hyperspectral images and Sentinel-2A MSI multispectral images. Hyperspectral: 1400×1400 pixels, 280 bands (400–2500 nm), 30 m resolution. Multispectral: 4200×4200 pixels, 4 bands, 10 m resolution.2. File StructureExperiment-Data/:Includes per-band PSNR values of CAVE, PaviaU, and WDCM datasets, used to plot the spectral performance curves in the paper.Experiment-Imgs/:Includes all experimental visualization results:Fusion result comparison images (4 figures)Per-band PSNR curve figures (3 figures)YRE-Dataset/:Real-world Yellow River Estuary dataset used in real-scene experiments.3. UsageThese data are used to reproduce all quantitative results, tables, and figures in the paper.All data are released underCreative Commons Attribution 4.0 International license.CRA-MHIF实验数据集：面向多光谱与高光谱图像融合的共识推理注意力网络 本套件包含对应论文《面向多光谱与高光谱图像融合的共识推理注意力网络》的全部配套数据。 1. 所用数据集 所有数据集均为公开可用，并严格按照论文所述方式使用。 CAVE数据集： 来源：https://cave.cs.columbia.edu/repository/Multispectral 描述：包含32个室内多光谱场景，具备31个光谱波段（波长范围400–700 nm，步长10 nm），空间分辨率为512×512。 帕维亚大学数据集（PaviaU）： 来源：https://www.ehu.eus/ccwintco/index.php/Hyperspectral_Remote_Sensing_Scenes#Pavia_Centre_and_University 描述：由意大利帕维亚上空的ROSIS-03传感器采集的高光谱图像，包含103个有效光谱波段（波长范围0.43–0.86 μm），空间分辨率为1.3 m。 华盛顿特区购物中心数据集（WDCM）： 来源：https://engineering.purdue.edu/~biehl/MultiSpec/hyperspectral.html 描述：由HYDICE机载传感器采集的高光谱图像，移除水汽吸收波段后剩余191个有效光谱波段，空间分辨率为2.5 m。 黄河口数据集（YRE）： 来源：YRE-Dataset/ 描述：真实世界遥感数据集，由高分五号（Gaofen-5）AHSI高光谱图像与Sentinel-2A MSI多光谱图像组成。其中高光谱图像分辨率为1400×1400像素，包含280个波段（波长范围400–2500 nm），空间分辨率30 m；多光谱图像分辨率为4200×4200像素，包含4个波段，空间分辨率10 m。 2. 文件结构 Experiment-Data/：包含CAVE、PaviaU及WDCM数据集的逐波段峰值信噪比（Peak Signal-to-Noise Ratio, PSNR）数据，用于绘制论文中的光谱性能曲线。 Experiment-Imgs/：包含全部实验可视化结果：融合结果对比图像（共4幅）、逐波段PSNR曲线图（共3幅）。 YRE-Dataset/：真实场景实验所用的黄河口真实遥感数据集。 3. 使用说明 本数据集可用于复现论文中的全部定量结果、表格与插图。所有数据采用知识共享署名4.0国际许可协议（Creative Commons Attribution 4.0 International License）发布。