青岛无人机高光谱基准数据集

本数据集（QUH）是针对高光谱图像精确土地覆盖分类构建的具有挑战性的无人机高光谱基准数据集。该数据集由中国石油大学（华东）孙根云教授团队发布。QUH数据集包含三个典型的子数据集，分别命名为QUH-Tangdaowan（唐岛湾）、QUH-Qingyun（庆云）和QUH-Pingan（平安）。这些数据采集自中国青岛的快速发展沿海区域，涵盖了复杂的异质地表场景，包括湿地、城市建成区和码头。数据集提供了高空间分辨率和高光谱分辨率的影像立方体及其对应的像素级地面实况标签。该数据集旨在解决高光谱分类中类内光谱变异和空间异质性的难题，可作为评估高光谱图像分类、降维及特征提取算法性能的标准基准数据。

This dataset (QUH) is a challenging UAV-borne hyperspectral benchmark dataset developed for precise land cover classification using hyperspectral images. It was released by the team led by Professor Genyun Sun from China University of Petroleum (East China). The QUH dataset comprises three typical sub-datasets, namely QUH-Tangdaowan (Tangdaowan), QUH-Qingyun (Qingyun), and QUH-Pingan (Pingan). All data were collected from the rapidly developing coastal region of Qingdao, China, covering complex heterogeneous surface scenarios including wetlands, urban built-up areas, and docks. The dataset provides hyperspectral image cubes with both high spatial and spectral resolution, alongside their corresponding pixel-level ground truth labels. This dataset is designed to address the challenges of intra-class spectral variability and spatial heterogeneity in hyperspectral classification, and can serve as a standard benchmark for evaluating the performance of hyperspectral image classification, dimensionality reduction, and feature extraction algorithms.

搜集汇总

背景与挑战

背景概述

青岛无人机高光谱基准数据集（QUH）是一个用于高光谱影像精确土地覆盖分类的挑战性基准数据集，包含三个子数据集，覆盖青岛典型区域，具有高空间和光谱分辨率。该数据集由中国石油大学（华东）孙根云教授团队发布，旨在解决高光谱分类中的类内光谱变异和空间异质性挑战。