基于步行GPS轨迹的路网提取研究

发布时间：2018-05-28 03:44

本文选题：智能交通系统 + 小路提取　；参考：《湖南科技大学》2014年硕士论文

【摘要】：智能交通系统有利于改善城市交通环境，合理分配交通资源，可以获取巨大的经济和社会效益。电子地图是构建智能交通系统的基础，自动生成电子地图和及时更新路网信息对于车辆导航、交通规划和土地利用等具有重要的应用价值和意义。目前大多基于遥感影像的生成方法只是针对某种类型的道路，即该方法的通用性导致了其准确性较差，而且影像更新周期较长使得路网信息难以及时更新。随着智能移动终端和移动互联网的快速发展，人们能够方便地获取GPS轨迹数据。GPS轨迹数据具有采集成本低、更新速度快和覆盖范围广等特点。因此，基于GPS轨迹的提取方法引起了学术界与产业界的广泛关注。目前，研究者通常基于浮动车或出租车的GPS轨迹来挖掘城市主干路网。但是，现有方法忽略了小路的自动提取，而这种小路保有量多且变更频繁，保障其完整、准确对于抗震救灾、小区导航或乡村游览等应用领域非常重要。针对该问题，本文提出基于步行GPS轨迹的小路提取方法。实验数据采用湖南科技大学的步行巡检GPS数据集，该数据集是巡检人员在巡逻过程中由随身携带的GPS记录仪所记录，共有1746万个轨迹点。本文方法主要包括三部分内容，分别为数据预处理、道路中心线生成和路网精度评价。首先，通过数据预处理去除原始GPS轨迹数据的异常值，确保数据的精确性。其次，利用GPS轨迹自动地生成道路中心线，，并进行路网的矢量化处理。最后，以百度地图等相关信息为参考路网，分别从定性与定量两方面对本文方法的路网提取进行评价。其中，道路中心线生成方法包括轨迹点聚类、聚类点分割和中心线拟合三部分。实验结果表明，本文方法能够准确提取路网，覆盖率可达96.21%，而误检率仅3.26%；并且能够提取小路和更新路网。
[Abstract]:Intelligent Transportation system (its) is beneficial to improve urban traffic environment, allocate traffic resources rationally, and obtain huge economic and social benefits. Electronic map is the basis of constructing intelligent transportation system. Automatic generation of electronic map and timely updating of road network information have important application value and significance for vehicle navigation, traffic planning and land use. At present, most of the methods based on remote sensing image are only for a certain type of road, that is, the generality of this method leads to its poor accuracy, and the long period of image updating makes it difficult to update the road network information in time. With the rapid development of intelligent mobile terminals and mobile Internet, it is easy to obtain GPS trajectory data. GPS trajectory data has the characteristics of low acquisition cost, fast update speed and wide coverage. Therefore, the extraction method based on GPS trajectory has attracted wide attention in academia and industry. At present, researchers usually mine the urban trunk road network based on the GPS tracks of floating cars or taxis. However, the existing methods ignore the automatic extraction of the path, which has a large number and changes frequently, and ensures its integrity, which is very important for the earthquake relief, community navigation or rural tourism and other application fields. To solve this problem, a path extraction method based on walking GPS locus is proposed in this paper. The experimental data are based on the walking GPS data set of Hunan University of Science and Technology. The data set is recorded by the GPS recorder carried by the patrol personnel during the patrol, with a total of 17.46 million locus points. This method includes three parts: data preprocessing, road centerline generation and road network accuracy evaluation. Firstly, the outliers of the original GPS trajectory data are removed by data preprocessing to ensure the accuracy of the data. Secondly, the road centerline is generated automatically by using GPS trajectory, and the vectorization of road network is carried out. Finally, taking Baidu map and other related information as reference road network, this paper evaluates the road network extraction from qualitative and quantitative aspects. Among them, the road centerline generation method includes trajectory point clustering, clustering point segmentation and centerline fitting. The experimental results show that the proposed method can extract the road network accurately, with a coverage rate of 96.21 and a false detection rate of only 3.26 percent, and can extract the path and renew the road network.
【学位授予单位】：湖南科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P228.4;U495

【参考文献】