| 胡振华,张乔艳.基于ASD与Hyperion高光谱数据的主要针叶树种分类研究[J].林业调查规划,2023,48(3):1-6 |
| 基于ASD与Hyperion高光谱数据的主要针叶树种分类研究 |
| Classification of Major Coniferous Tree Species Based on ASD and Hyperion Hyperspectral Data |
| |
| DOI： |
| 中文关键词: 树种识别分类 ASD数据 Hyperion数据 特征波段选择 支持向量机 |
| 英文关键词: forest species identification ASD data Hyperion data feature band selection support vector machine |
| 基金项目:贵州省林业科研项目（黔林科合J字〔2022〕21号；黔林科合J字〔2021〕01号）. |
|
| 摘要点击次数: 64 |
| 全文下载次数: 60 |
| 中文摘要: |
| 以云南省香格里拉市为研究区，对ASD光谱仪实测的4种针叶树种光谱数据采用包络线去除法、光谱一阶微分法和光谱二阶微分法3种波段选择方法得到Hyperion高光谱影像数据的分类特征波段，采用最大似然法、支持向量机2种分类方法对所选的特征波段开展树种识别分类，对原始影像采用光谱角填图分类方法作对比实验。结果表明，基于ASD数据的光谱一阶波段选择方案的支持向量机分类方法精度最高，总体分类精度为81.95%，Kappa系数为0.725 1。采用ASD实测光谱数据能有效指导Hyperion进行树种分类，基于数据尺度和换算方式，一阶微分更适合特征波段选择；与传统的数理统计分类方法和光谱特征分类方法相比，基于机器学习的方法如支持向量机等在高光谱遥感分类中具有更大的应用潜力。 |
| 英文摘要: |
| Taking Shangri-La City, Yunnan Province as the study area, the spectral data of four coniferous species measured by ASD spectrometer were selected by three band selection methods of continuum removal method, spectral first-order differentiation method and spectral second-order differentiation method to obtain the classification characteristic bands of Hyperion data. The classification methods of maximum likelihood method and support vector machine were used to identify and classify the selected feature bands, and the classification method of spectral angle mapping was used to compare the original image. The results showed that the support vector machine classification method based on ASD data for the first-order spectral band selection scheme had the highest accuracy, with the overall classification accuracy of 81.95% and Kappa coefficient of 0.725 1. The measured spectral data of ASD could effectively guide Hyperion to classify tree species, and the first-order differentiation was more suitable for feature band selection due to data scale and conversion method; compared with traditional mathematical statistics classification methods and spectral feature classification methods, machine learning-based methods such as support vector machine had greater application potential in hyper spectral remote sensing classification. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
