摘要
本文基于随机森林算法利用多时序Landsat8影像多波段作为分类特征,通过调整随机森林模型参数,对不同模型进行分类精度评估,实现了遥感影像土地利用分类。研究结果表明,基于混淆矩阵、分类准确率、Kappa系数等分类评价指标,随机森林算法获得了较好的地物分类结果。在模型参数方面,决策树数目对分类精度具有较大影响,通过调整参数得到最优随机森林模型。当决策树数目为25时,分类正确率达到97%以上,Kappa系数达到0.96。
In this paper,based on the random forest algorithm,the land use classification of remote sensing images is achieved by using multi-band of multi-temporal Landsat8 images as classification features.The classification accuracy of different models are evaluated by adjusting the parameters of the random forest model.The research results show that based on the classification e valuation indexes such as confusion matrix,classification accuracy,and Kappa coefficient,the random forest algorithm obtains better land feature classification results.In terms of model parameters,the number of decision trees has a large influence on the classification accuracy,and the optimal random forest model is obtained by adjusting the parameters.As a result,the classification accuracy reaches over 97%when the number is taken as 25,and the Kappa coefficient r eaches 0.96.
作者
王煜炜
WANG Yuwei(School of Artificial Intelligence,Jianghan University,Wuhan Hubei 430056,China)
出处
《信息与电脑》
2022年第15期82-84,共3页
Information & Computer
基金
江汉大学2019年高层次人才科研启动经费项目(项目编号:1028/06070001)。
关键词
机器学习
随机森林
遥感影像
土地利用分类
machine learning
random forest
remote sensing image
land use classification