Hyperspectral Imaging Honey Database

This a spreadsheet of features extracted from hyperspectral images of New Zealand honey. The images of honey have been captured with a hyperspectral imager SOC-710 from Surface Optic. The hypercubes captured are between 400 - 1000nm in wavelength with a 5nm increment and have a 520 x 696 spatial resolution.The system used to capture these images has been developed using rigorous testing and evaluation to create a standard for the hyperspectral imagery capture of honey.The dataset comprises the features (128 spectral bands) of the hyperspectral images after a well-designed procedure for calibration, pre-processing and segmentation. This process produces a consistent dataset through all trials, and so it is comparable to the initial work on feature reduction and classification. There are six acquisitions of each honey sample; these are then captured by the hyperspectral imagery and calibrated, pre-processed and segmented to get the final spectral information.本数据集为从新西兰蜂蜜高光谱图像（hyperspectral image）中提取的特征电子表格。本次蜂蜜图像采用Surface Optic公司生产的SOC-710型高光谱成像仪（hyperspectral imager）采集，所获取的高光谱数据立方体（hypercube）波长范围为400~1000nm，光谱采样步长为5nm，空间分辨率为520×696。本次图像采集系统经过严格的测试与评估开发而成，旨在建立蜂蜜高光谱图像采集的标准规范。本数据集包含经过精心设计的校准、预处理与分割流程处理后的高光谱图像特征，共包含128个光谱波段。该流程可在所有试验中生成一致性统一的数据集，因此可与此前的特征降维与分类相关研究成果进行对比。每个蜂蜜样本均进行6次图像采集；采集得到的图像经校准、预处理与分割流程处理后，即可得到最终的光谱信息。