低频轨迹数据的多时相增量式道路提取方法  

A method of multi-temporal incremental extraction of road network based on low-frequency trajectory data

作　　者：张云菲 佘婷婷[2] 邓敏[2] 周访滨[1] ZHANG Yunfei;SHE Tingting;DENG Min;ZHOU Fangbin(School of Traffic&Transportation Engineering,Changsha University of Science&Technology,Changsha 410014,China;School of Geosciences and Info-physics,Central South University,Changsha 410083,China)

机构地区：[1]长沙理工大学交通运输工程学院,长沙410114 [2]中南大学地球科学与信息物理学院,长沙410083

出　　处：《测绘科学》2020年第4期192-198,共7页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(41601495,41971421,41730105,41771492,41671446);湖南省自然科学基金项目(2018JJ3525);湖南省教育厅科学研究项目(18C0228);测绘遥感信息工程国家重点实验开放基金项目(17s01);长沙理工大学公路地质灾变预警空间信息技术湖南省工程实验室开放基金项目(kfj170605)。

摘　　要：针对目前利用低频时空轨迹数据进行道路提取与更新,难以同时满足提取精度和算法效率要求的问题,该文提出一种基于低频轨迹数据的多时相增量式道路提取方法。首先,将原始轨迹数据分割为多个时相轨迹序列,并利用栅格化法得到多个时相的初始道路中心线;接着,采用吸引力模型纠正初始道路中心线的位置偏差,再通过k-segment主曲线算法拟合得到最终道路中心线,构建道路骨架地图;最后,统计与道路中心线关联的轨迹位置与方向信息,进一步挖掘道路宽度与单双向通行规则等交通语义信息,丰富道路骨架地图。实验结果表明,基于多时相轨迹分别提取与增量融合的方式,可有效发现道路网的时空变化规律,获得较高的道路检测概率与信息提取准确率,同时避免对全时段轨迹直接处理的复杂运算量。Aiming at the problem that presently,there are a recent surge in road information extraction and road map updating based on low-frequency trajectory data,but previous methods can hardly satisfy the balance of extraction precision and time efficiency. The paper presented a multi-temporal incremental method for extracting road network from low-frequency trajectory data. The proposed method adopted a combined strategy of multi-temporal extraction and incremental fusion. Firstly,the raw trajectory data was partitioned into several multi-temporal track sequences and multi-temporal road networks were roughly extracted from those multi-temporal track sequences by a rasterization method. Secondly,the extracted multi-temporal road networks were adjusted by attraction forced-based model and the road centerlines were then fitted to construct a road skeleton map by k-segment principle curve algorithm. Finally,the road skeleton map was further semantically enriched by inferring the road width and traffic rule(e.g.,single way or double way) according to the positional distribution and travelling direction of associated trajectories. The experimental results showed that the proposed method could efficiently discover the time-space changing law of road network,achieve a high precision of road detection and information extraction,and meanwhile avoid complex computation of full-time trajectory data processing.

关 键 词：道路提取 GPS轨迹 多时相分割 增量融合 

分 类 号：P208[天文地球—地图制图学与地理信息工程]