出租车GPS大数据的道路行车可视分析

Visual Analytics of Road Traffic with Large Scale Taxi GPS Data

针对出租车GPS数据因其数据量庞大和时空信息特征复杂而带来的分析难题,提出一种基于出租车GPS大数据的道路行车可视分析方法.该方法用OpenStreetMap得到开阔道路的地图,采用离散和连续型2种编码方式对道路上的车流量、行车方向和速度等情况进行分析;离散编码采用箭头图表示,并用速度区间聚类算法优化颜色布局;连续编码采用栈图表示,并用特征点提取算法加速图表绘制.最后以杭州市出租车GPS数据为样例,将数据分布式存储在云计算平台上,采用MapReduce加快数据查询和处理,应用文中的2种可视编码方式进行可视分析,结果表明,该方法能准确地反映杭州市道路交通状况.

With large-scale and complicated spatio-temporal characteristics, visual analytics of taxi GPS data is a challenging issue. In this paper, we present a visual analytic method for road traffic analysis based on taxi GPS data, and we adopt two encoding schemes, the discrete arrow graph and the continuous stack graph, to explore the volume, direction, speed and other information of road traffic flow on widened roads based on OpenStreetMap. Douglas-Peucker algorithm and the velocity clustering algorithm are used for data reduction and improving rendering respectively. The preprocessed taxi GPS data are stored in a cloud computing platform in a distributed manner, and MapReduce is utilized to accelerate data and query processing. We test the validities of our proposed encoding schemes on Hangzhou taxi GPS data. Experimental results show that our method can effectively and accurately reveal the status of road traffic in Hangzhou.