海量出租车轨迹数据探索性分析方法的研究与实现
发布时间：2018-08-07 17:10
【摘要】：随着卫星技术、无线通讯技术和定位技术的迅猛发展,现在已可以快速、便捷地获取海量的车辆轨迹数据。这些轨迹数据具有数量多、覆盖广、密度大的特点,对其进行分析和挖掘,可以获得车辆关于移动过程方面的信息。要应用这些海量轨迹数据,需首先对其进行探索性分析,而该分析的重要基础就是数据的可视化。车辆轨迹数据从其物理构成上来说,由大量矢量轨迹点组成。现有成熟商业数据库管理软件内部采用的可视化机理是通过创建矢量索引的方式来实施可视化。这种方法的缺点在于,当点的数量激增时,如数百万个点,可视化的效率会迅速降低,也阻碍进一步的探索性分析。为了解决此问题,本文提出了一种基于影像金字塔的可视化方法,并在此基础上展开探索性分析。 影像金字塔是指在同一空间参照下,由分辨率从高到低、数据量从大到小的图片构成的金字塔状影像集,是一种适用于栅格图像的分层数据结构形式,可以满足不同的显示要求。我们提出的可视化方法简述如下：将矢量的轨迹点转换成栅格数据,平面上,为其建立格网索引,纵向上,为其构建影像金字塔。为避开硬件限制,防止图片太大无法生成,以及提高存取效率,将金字塔每层图片再分割成地图瓦片。显示时,根据当前显示级别,将视图范围与金字塔的某层进行求交运算,相交部分的图片或图片的一部分画到屏幕显示。文中详细介绍了矢量轨迹点到栅格数据的转换,格网索引的建立,影像金字塔及地图瓦片的构建方法,以及可视化操作的实现步骤,并实现了一个案例应用。以上海市某日出租车轨迹和深圳市多日出租车轨迹作为实验数据,为分析不同参数对可视化效率的影响,共设计了四组对比实验,通过运行本文所提出的可视化方法,实现了轨迹的快速可视化,运行结果显示,本文提出的方法可以有效地解决海量车辆轨迹数据可视化效率低下的问题。在实现可视化方法的基础上开展了初步的探索性分析,包括轨迹信息回放、行驶信息统计和驾驶行为分析等功能。行驶信息统计包括计算载客时间、空载时间和平均速度等；驾驶行为分析指计算出租车在载客状态下的超速、急加速和急减速比率。
[Abstract]:With the rapid development of satellite technology, wireless communication technology and positioning technology, massive vehicle trajectory data can be obtained quickly and conveniently. These trajectory data have the characteristics of large quantity, wide coverage and high density. By analyzing and mining them, we can obtain the information about the moving process of the vehicle. In order to apply these massive trajectory data, it is necessary to make an exploratory analysis of them first, and the important foundation of this analysis is the visualization of the data. Vehicle trajectory data is composed of a large number of vector locus points in terms of its physical composition. The visualization mechanism used in existing commercial database management software is to create vector index to implement visualization. The disadvantage of this method is that when the number of points increases, millions of points, the efficiency of visualization will be reduced rapidly, and the further exploratory analysis will be hindered. In order to solve this problem, a visualization method based on image pyramid is proposed, and an exploratory analysis is carried out. Image pyramid is a kind of pyramid image set which is composed of images with high resolution from high to low and data from large to small under the same spatial reference. It is a kind of layered data structure suitable for raster images. Can meet different display requirements. The visualization method proposed by us is as follows: the vector locus points are converted into raster data, the grid index is built on the plane, and the image pyramid is constructed longitudinally. To avoid hardware constraints, prevent images from being too large to generate, and improve access efficiency, each layer of the pyramid is further divided into map tiles. Display, according to the current level of display, the scope of the view and a layer of the pyramid to calculate the intersection, the intersection part of the picture or part of the picture to the screen display. In this paper, the transformation of vector locus to raster data, the establishment of grid index, the construction of image pyramid and map tile, and the realization of visual operation are introduced in detail, and a case application is realized. In order to analyze the effect of different parameters on the visualization efficiency, four groups of comparative experiments were designed based on the experimental data of one day taxi track in Shanghai and the multi-day taxi track in Shenzhen City, and the visualization method proposed in this paper was run. The results show that the method proposed in this paper can effectively solve the problem of low visualization efficiency of massive vehicle trajectory data. Based on the visualization method, the preliminary exploratory analysis is carried out, including the functions of track information playback, driving information statistics and driving behavior analysis. The statistics of driving information include the calculation of passenger time, no-load time and average speed, and the analysis of driving behavior refers to the calculation of the ratio of speeding, rapid acceleration and rapid deceleration of taxis under the condition of carrying passengers.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P208;U495
本文编号：2170754
[Abstract]:With the rapid development of satellite technology, wireless communication technology and positioning technology, massive vehicle trajectory data can be obtained quickly and conveniently. These trajectory data have the characteristics of large quantity, wide coverage and high density. By analyzing and mining them, we can obtain the information about the moving process of the vehicle. In order to apply these massive trajectory data, it is necessary to make an exploratory analysis of them first, and the important foundation of this analysis is the visualization of the data. Vehicle trajectory data is composed of a large number of vector locus points in terms of its physical composition. The visualization mechanism used in existing commercial database management software is to create vector index to implement visualization. The disadvantage of this method is that when the number of points increases, millions of points, the efficiency of visualization will be reduced rapidly, and the further exploratory analysis will be hindered. In order to solve this problem, a visualization method based on image pyramid is proposed, and an exploratory analysis is carried out. Image pyramid is a kind of pyramid image set which is composed of images with high resolution from high to low and data from large to small under the same spatial reference. It is a kind of layered data structure suitable for raster images. Can meet different display requirements. The visualization method proposed by us is as follows: the vector locus points are converted into raster data, the grid index is built on the plane, and the image pyramid is constructed longitudinally. To avoid hardware constraints, prevent images from being too large to generate, and improve access efficiency, each layer of the pyramid is further divided into map tiles. Display, according to the current level of display, the scope of the view and a layer of the pyramid to calculate the intersection, the intersection part of the picture or part of the picture to the screen display. In this paper, the transformation of vector locus to raster data, the establishment of grid index, the construction of image pyramid and map tile, and the realization of visual operation are introduced in detail, and a case application is realized. In order to analyze the effect of different parameters on the visualization efficiency, four groups of comparative experiments were designed based on the experimental data of one day taxi track in Shanghai and the multi-day taxi track in Shenzhen City, and the visualization method proposed in this paper was run. The results show that the method proposed in this paper can effectively solve the problem of low visualization efficiency of massive vehicle trajectory data. Based on the visualization method, the preliminary exploratory analysis is carried out, including the functions of track information playback, driving information statistics and driving behavior analysis. The statistics of driving information include the calculation of passenger time, no-load time and average speed, and the analysis of driving behavior refers to the calculation of the ratio of speeding, rapid acceleration and rapid deceleration of taxis under the condition of carrying passengers.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P208;U495
【引证文献】
相关硕士学位论文 前2条
1 赵苗苗;基于出租车轨迹数据挖掘的推荐模型研究[D];首都经济贸易大学;2015年
2 钱科宇;基于WebGIS的车辆监控系统的性能优化研究[D];南京邮电大学;2015年
，本文编号：2170754
本文链接：https://www.wllwen.com/kejilunwen/dizhicehuilunwen/2170754.html
