基于BP神经网络和拓扑参数的道路网选取研究

发布时间：2018-01-03 12:25

本文关键词：基于BP神经网络和拓扑参数的道路网选取研究　出处：《测绘科学技术学报》2016年03期 　论文类型：期刊论文

【摘要】：道路网选取是自动制图综合的重点和难点之一,运用智能化方法实现选取是当前研究的热点。BP神经网络具有强大的非线性映射能力,可以模仿人脑机能,通过对样本的学习和训练实现自动选取;结合拓扑参数,可以使选取结果很好地保持原道路网的连通性和整体结构特征。因此,提出一种基于BP神经网络和拓扑参数的道路网选取方法。首先选择训练样本并计算其拓扑参数;然后设计BP神经网络的结构,利用训练样本进行训练,找出最佳网络结构;最后选取不同特征的道路网进行实验,将选取结果与专家选取的结果进行对比分析,评价了该方法的优势与不足,并指出了下一步的改进方向。
[Abstract]:Road network selection is one of the key and difficult points in automatic cartographic generalization. Using intelligent method to realize selection is a hot topic in current research. BP neural network has powerful nonlinear mapping ability and can imitate the function of human brain. Through the study and training of the samples, the automatic selection is realized. Combined with topological parameters, the selection results can keep the connectivity and overall structural characteristics of the original road network very well. A road network selection method based on BP neural network and topological parameters is proposed. Firstly, the training samples are selected and their topological parameters are calculated. Then the structure of BP neural network is designed and the best network structure is found by using training samples. Finally, the road network with different characteristics is selected for experiments. The results are compared with those of experts, the advantages and disadvantages of the method are evaluated, and the next improvement direction is pointed out.
【作者单位】：南京大学地理信息科学系;南京师范大学虚拟地理环境教育部重点实验室;南京师范大学地理科学学院;
【基金】：国家自然科学基金项目(41371433)
【分类号】：P283.7
【正文快照】： 道路是地图上重要的地物,道路网选取是地图自动综合研究的重点和难点之一[1-2],也是至今没有完全解决的问题,众多地图学者做了大量研究[3-6]。究其原因,道路网比较复杂,需要考虑的因素和指标比较多,而且这些指标与道路重要性的关系比较模糊,没有明确的数学模型和公式描述,这就

【相似文献】