基于改进随机森林算法的国土空间规划用地分类研究  

Study on Land Use Classification of Land Spatial Planning Based on Improved Random Forest Algorithm

作　　者：魏国杰 穆战强 朱飞 Wei Guojie;Mu Zhanqiang;Zhu Fei(Henan Urban and Rural Planning,Design and Research Institute Co.,Ltd,Zhengzhou 450000,China)

机构地区：[1]河南省城乡规划设计研究总院股份有限公司,郑州450000

出　　处：《科技通报》2022年第10期11-15,28,共6页Bulletin of Science and Technology

摘　　要：为了提升国土空间规划用地分类效果,以满足国土空间管控需求,提出基于改进随机森林算法的国土空间规划用地分类方法。在已有国土空间规划用地分类标准基础上,有效结合现有国土空间管控需求,制定国土空间规划用地分类标准。在此基础上,采用遥感技术获取国土空间规划用地遥感图像,提取遥感图像颜色、形状与纹理特征,通过LDA算法处理、简化遥感图像特征数据,以处理好的数据为依据采用AdaBoost改进随机森林算法,得到最佳的国土空间规划用地分类器,实现了国土空间规划用地精准分类。实验数据显示,提出方法的分类精度、灵敏度与特异度数值更大,证实了该方法具备较好的分类效果,能够充分满足国土空间管控需求。In order to improve the land classification effect of land spatial planning and meet the needs of land spatial management and control,a land classification method of land spatial planning based on improved random forest is proposed.Based on the existing land classification standards for land space planning,and effectively combined with the existing land space management and control requirements,formulate land classification standards for land space planning.On this basis,the remote sensing image of land for land spatial planning is obtained by using remote sensing technology,and the color,shape and texture features of the remote sensing image are extracted.The feature data of the remote sensing image is processed and simplified by LDA Algorithm.Based on the processed data,the best land classifier for land for land spatial planning is obtained by using AdaBoost improved random forest algorithm,it has realized the accurate classification of land for land spatial planning.The experimental data show that the classification accuracy,sensitivity and specificity of the proposed method are greater,which proves that the method has good classification effect and can fully meet the needs of land and space management and control.

关 键 词：改进随机森林算法 国土空间规划 规划用地分类 遥感技术 LDA算法 

分 类 号：P902[天文地球—自然地理学] TU98[建筑科学—城市规划与设计]