BJTU-UVA Dataset
收藏资源简介:
BJTU-UVA数据集是由北京交通大学和阿姆斯特丹大学合作创建的,专门用于高光谱图像(HSI)自动光谱校准的首个大型数据集。该数据集包含765对高质量的高光谱图像对,覆盖了多样化的自然场景和光照条件。数据集的创建过程采用了异步校准方法,使用204个波段的高光谱相机进行采集,空间分辨率为512×512,量化位数为12位。该数据集不仅用于校准研究,还扩展了10种不同的物理测量光照条件,生成了7650对未校准的高光谱图像。BJTU-UVA数据集主要应用于高光谱图像的自动校准,旨在解决传统物理参考校准方法中的遮挡、手动操作和相机移动限制等问题。
The BJTU-UVA dataset, co-created by Beijing Jiaotong University and the University of Amsterdam, is the first large-scale dataset specifically developed for automated spectral calibration of hyperspectral images (HSI). It contains 765 pairs of high-quality hyperspectral images covering diverse natural scenes and lighting conditions. During its construction, an asynchronous calibration method was adopted, with data collected using a hyperspectral camera equipped with 204 spectral bands, a spatial resolution of 512×512, and a 12-bit quantization depth. Beyond serving calibration research, the dataset further generates 7,650 pairs of uncalibrated hyperspectral images by incorporating 10 distinct physically measured lighting conditions. Primarily applied to automated hyperspectral image calibration, this dataset aims to resolve issues including occlusion, manual operation, and camera movement constraints in traditional physical reference calibration methods.




