一种面向相似查询的轨迹索引方法  

A METHOD OF TRACK INDEX FOR SIMILARITY SEARCH

作　　者：王飞[1] 庞悦 周向东[1] 陈海波 

机构地区：[1]复旦大学计算机科学技术学院,上海200433 [2]国网上海市电力公司,上海200122

出　　处：《计算机应用与软件》2017年第11期1-5,63,共6页Computer Applications and Software

基　　金：国家高技术研究发展计划项目(2015AA050203);国家自然科学基金项目(61370157);国家电网公司总部科技项目(52094016000A)

摘　　要：轨迹数据具有重要的应用价值,轨迹索引技术得到广泛的研究与关注。传统索引方法存在节点重叠、缺乏动态划分空间能力和丢失大量原始信息等问题,为此提出一种面向相似查询的轨迹索引方法GeoSAX。该方法将原始轨迹分成若干等长子段并采用基于Geohash的空间编码;对编码后的整条轨迹设计了基于HBase存储的索引架构;实现相似轨迹查询。GeoSAX不仅节点间没有重叠,还能依据数据量的大小对空间动态划分,同时保留指定精度的轨迹信息。在真实的航运和出租车数据集上进行的对比实验表明,与传统方法相比GeoSAX具有更好的轨迹查询性能。Because the tracking data have important application value, the track index technology has been widely studied and concerned. Traditional indexing methods have many problems such as node overlap, lack of dynamic partitioning of spatial capabilities and loss of a large number of original information. Therefore, we propose GeoSAX, a track index method for similarity search. In this method, the original tracking was divided into several equal segments, and spatial coding based on Geohash was adopted. We designed an indexing architecture for the whole track after encoding, which was based on HBase storage. Thus, the similarity search was realized. GeoSAX not only does not overlap between nodes, but also dynamically divides the space according to the size of the data, while preserving the track information of the specified precision. Contrast experiments on real shipping and taxi data sets show that GeoSAX has better track search performance than traditional methods.

关 键 词：轨迹索引 相似查询 Geohash 空间编码 HBASE 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]