摘要
道路多级选取是地图综合的核心问题。针对现有方法多局限于小范围区域的问题,创新性地提出融合自适应特征与多级分类一体化的城市全域道路选取方法。通过多属性约束将路网划分为stroke单元,构建自适应空间特征和语义特征,结合GraphSAGE深度学习模型实现多级分类训练。实验表明,该方法在城市全域道路选取中实现了较高选取精度,并具有良好的跨区域泛用性。
Multi-level road selection is a core issue in map generalization.To address the limitations of existing methods that are often confined to small-scale areas,we innovatively proposed an urban road selection method integrating adaptive features and multi-level classification.We segmented the road network into stroke units through multi-attribute constraints,constructed adaptive spatial and semantic features,and leveraged the GraphSAGE deep learning model to achieve multi-level classification training.Experimental result demonstrates that the proposed method achieves high selection accuracy in urban road selection and has strong cross-region generalizability.
作者
罗浩
廖顺华
周何帆
程灵雅
邓维熙
张璇
LUO Hao;LIAO Shunhua;ZHOU Hefan;CHENG Lingya;DEN Weixi;ZHANG Xuan(The Third Geoinformation Mapping Institute of Ministry of Natural Resource,Chengdu 610100,China;Guangxi Zhuang Autonomous Region Institute of Cartography,Nanning 530200,China;Key Laboratory of Digital Mapping and Land Information Application,Ministry of Natural Resources,Chengdu 610100,China)
出处
《地理空间信息》
2025年第6期121-125,共5页
Geospatial Information
