道路网多特征匹配优化算法

发布时间：2018-05-05 05:41

本文选题：道路网匹配 + 支持向量机　；参考：《测绘学报》2016年05期

【摘要】：同名道路匹配技术是道路数据集成、更新和融合的重要前提。道路网匹配在智能交通(intelligent transportation system,ITS)与位置服务(location-based service,LBS)等方面具有重要的研究价值和应用意义。本文提出了一种道路网多特征匹配优化算法:首先从形状、距离、语义3方面分别设计了基于面积累积的形状差、综合中值Hausdorff距离和全局加权属性项距离3种相似性度量,以更准确地描述道路待匹配对之间的特征差异;然后通过SVM对相似性特征样本集训练,以构建道路网回归匹配模型;最后利用此模型对未知匹配结果道路待匹配对进行匹配结果预测。大量试验结果表明,本文算法对非线性偏差明显的道路网数据能够实现较高的匹配准确率和召回率,能有效地用于包含多重匹配关系的道路网匹配。
[Abstract]:Road matching with the same name is an important prerequisite for road data integration, updating and integration. Road network matching has important research value and application significance in intelligent transportation system (ITS) and location service (location-based service). In this paper, a multi-feature matching algorithm for road network is proposed. Firstly, three kinds of similarity measures are designed based on shape, distance and semantics, which are based on area cumulative shape difference, comprehensive median Hausdorff distance and global weighted attribute distance, respectively. In order to describe the feature difference between the road matching pairs more accurately, then train the similarity feature sample set by SVM to construct the road network regression matching model. Finally, the model is used to predict the matching results of unknown matching results. A large number of experimental results show that the proposed algorithm can achieve high matching accuracy and recall rate for road network data with obvious nonlinear deviations and can be effectively used for road network matching with multiple matching relationships.
【作者单位】：武汉大学遥感信息工程学院;地球空间信息技术协同创新中心;长江大学地球科学学院;
【基金】：国家自然科学基金(41561084;41201409;41201395) 山东省自然科学基金(ZR2014DL001)~~
【分类号】：U495

【相似文献】