应用GF-6遥感数据识别道路材质方法实验数据集

利用GF-6遥感数据可以识别道路材质，重要环节是数据识别模型的建立和道路不同材质的识别指标。作者以河北省廊坊部分地域作为实验区，采用计算光谱特征指数——差值指数、比值指数、方差指数和归一化指数，对不同道路材质的光谱特性进行分析；然后基于Google Earth影像和百度街景数据收集道路材质类型样本，利用机器学习技术，研发出应用GF-6遥感数据识别道路材质方法实验数据集。本次实验结果道路材质识别精度达到80.07%，Kappa系数为0.70。该实验数据集由3部分组成：（1）光谱特征指数数据；（2）道路材质样本数据；（3）道路材质识别结果数据。数据集存储为.dat、.shp.和.xlsx格式，由16个数据文件组成，数据量为3.69 GB（压缩为4个文件，1.62 GB）。

Road material identification can be achieved using GF-6 remote sensing data, with the core steps involving the establishment of data recognition models and the development of recognition indicators for different road materials. The authors selected a partial area of Langfang City, Hebei Province as the experimental study area, and calculated multiple spectral feature indices including Difference Index, Ratio Index, Variance Index, and Normalized Index to analyze the spectral characteristics of various road materials. Subsequently, road material type samples were collected using Google Earth imagery and Baidu Street View data, and an experimental dataset for road material identification with GF-6 remote sensing data was developed through machine learning techniques. The experimental results demonstrate that the road material identification accuracy reaches 80.07%, with a Kappa coefficient of 0.70. This experimental dataset comprises three components: (1) Spectral feature index data; (2) Road material sample data; (3) Road material identification result data. The dataset is stored in .dat, .shp, and .xlsx formats, consisting of 16 individual data files with a total size of 3.69 GB (compressed into 4 files with a total size of 1.62 GB).

搜集汇总

背景与挑战

背景概述

该数据集是基于GF-6遥感数据，通过机器学习方法识别道路材质的实验数据集，覆盖2020年河北省廊坊市区域。数据集包含光谱特征指数、道路材质样本和识别结果三部分，总数据量3.69 GB，存储为.dat、.shp和.xlsx格式，识别准确率达80.07%，Kappa系数0.70，适用于大规模道路材质识别应用。