基于Sentinel-2多光谱遥感影像的森林沼泽信息提取及精度评价

Forest Swamp Information Extraction and Accuracy Evaluation Based on Multispectral Remote Sensing Images of Sentinel-2

本文基于Sentinel-2多光谱遥感影像数据,使用监督分类方法、随机森林方法和卷积神经网络方法对江西省森林沼泽湿地进行信息提取。同时,使用混淆矩阵(Confusion Matrix)来判定三种遥感影像数据分类方法的精度。结果表明,三种遥感影像信息提取方法中,监督分类方法的精度相对较差、湿地分类混淆情况较为严重,其次为随机森林方法,卷积神经网络方法最优。本研究能够为地区湿地信息提取与遥感分类提供科研分析依据,为湿地保护和资源开发提供数据支持。

Based on Sentinel-2 multispectral remote sensing image data,this article uses supervised classification method,random forest method and convolutional neural network method to extract information from forest swamp wetlands in Jiangxi Province.At the same time,the Confusion Matrix is used to determine the accuracy of the three remote sensing image data classification methods.The results show that among the three remote sensing image information extraction methods,the accuracy of the supervised classification method is relatively poor,and the confusion of wetland classification is more serious,followed by the random forest method,and the convolutional neural network method is the best.The results is able to provide scientific research and analysis basis for regional wetland information extraction and remote sensing classification,and provide data support for wetland conservation and resource development.