面向土地交通整体规划的土地利用精细分类模型研究

Study on High-resolution Classification Model of Land Use for Overall Planning of Land Transportation

土地利用是交通需求的根源,精细的土地利用信息能够提高土地交通整体规划模型中社会经济活动的空间分布精度,从而为交通规划模型提供可靠的出行OD,人口、就业空间分布等信息.为获得精细的土地利用信息,以建筑物为研究单元,开展面向土地交通整体规划的土地利用精细分类模型研究.在分析土地利用与交通之间的关系后,建立城市土地利用分类体系;利用建筑物数据构建特征向量;基于Stacking思想对决策树、SVM、随机森林模型进行融合,得到分类结果,并与单个模型进行比较,评价模型的精度;对分类结果进行空间位置关系分析,以提高分类精度.最后以武汉市江岸区作为研究对象,进行分类算法的实验.结果显示,基于Stacking思想的融合模型的分类精度可以达到0.80,决策树、SVM、随机森林的分类精度分别为0.61、0.65、0.71;对融合模型的分类结果进行空间位置关系分析后,分类精度达到0.83.故构建的融合模型可以有效地识别出建筑物的土地利用类型,实现了城市土地利用的精细分类.

To obtain high resolution land use information,this paper focuses on fine classification of land use at building level.After analyzing the relationship between land use and traffic,the classification system of urban land use was established,and the feature vector was constructed by building data.Based on the idea of Stacking,the decision tree,SVM and random forest model were fused,and the classification results were obtained and compared with a single model to evaluate the accuracy of the model.The spatial position relationship of classification results was analyzed to improve the classification accuracy.Finally,taking Jiang’an District of Wuhan as the research object,the classification algorithm was tested.The results show that the classification accuracy of the fusion model based on the idea of Stacking can reach 0.80,and the classification accuracy of decision tree,SVM and random forest are 0.61,0.65 and 0.71 respectively.After analyzing the spatial position relationship of the classification results of the fusion model,the classification accuracy reaches 0.83.Therefore,the fusion model can effectively identify the land use types of buildings and realize the fine classification of urban land use.