【摘要】：當前,我國城市道路建設速度遠遠落后于汽車數量增長的速度,再加上交通資源配置不合理,造成了經常性的交通擁堵,導致嚴重的資源浪費和環(huán)境污染。智能交通系統(tǒng)作為一種新型運輸系統(tǒng),可以合理分配資源、提高道路通行能力,有效地改善交通狀況。電子地圖是智能交通系統(tǒng)的最基本元素,其基本功能包括路網顯示、實時導航、路徑推薦等等,因其信息豐富、交互性好等優(yōu)點,極大地方便了人們的出行。但是,傳統(tǒng)的電子地圖制作方法成本高、更新慢,難以適應當前路網更新頻繁的現(xiàn)狀,因此,研究如何快速有效地更新路網信息,從而為人們提供最新的導航、路徑推薦等功能,具有很強的現(xiàn)實意義和廣闊的應用前景。GPS數據記載了移動對象的軌跡信息,如果對這些數據加以分析,就可以為路網更新提供輔助信息,節(jié)省電子地圖的制作成本。此外,帶有定位功能的硬件設備價格降低和智能手機的迅速普及,使得人們可以很容易收集到道路GPS數據。因此,基于GPS軌跡的時空數據挖掘技術逐漸成為了智能交通系統(tǒng)領域的研究熱點之一。出租車作為商業(yè)性的交通工具,具有行駛時間長、軌跡范圍廣、車輛基數大、數據收集成本低的優(yōu)點,因此,使用出租車GPS軌跡數據進行路網更新及路徑選擇具有非常大的優(yōu)勢。本文采用微軟亞洲研究院(MSRA)提供的公開數據集(2007年4月到2010年10月的北京市出租車GPS數據),主要做了以下三個方面的工作:(1)GPS軌跡預處理。首先去除帶有誤差的數據,再利用滾動式聚類算法得到所有軌跡數據區(qū)域內主干道的道路,最后通過與Google Map對比,證明了路網的準確性;(2)提出一種基于出租車GPS軌跡的路網拓撲構建方法。它是路徑導航和道路選擇的前提。為了獲得最新的道路及其連接關系,本文采用基于出租車GPS軌跡的路口識別方法,利用(1)中獲得的道路,考慮路口處道路點數目要多的特征提取出路口,并進一步完成路網的拓撲構建,最后通過與Google Map、北京市shp地圖對比,證明了路口的準確性;(3)提出了一種基于四叉堆的道路選擇方法。根據(1)中的道路以及(2)中獲得的拓撲關系和各條道路段長度,結合歷史GPS軌跡數據,使用一個基于四叉堆優(yōu)先隊列的Dijkstra方法進一步實現(xiàn)了最短、最優(yōu)路徑選擇的功能。
[Abstract]:At present, the speed of urban road construction in our country lags far behind the increasing speed of the number of cars, coupled with the unreasonable allocation of traffic resources, resulting in frequent traffic congestion, resulting in serious waste of resources and environmental pollution. As a new type of transportation system, intelligent transportation system can allocate resources reasonably, improve traffic capacity and improve traffic condition effectively. Electronic map is the most basic element of intelligent transportation system. Its basic functions include road network display, real-time navigation, path recommendation and so on. Because of its rich information, good interaction and other advantages, it greatly facilitates people to travel. However, the traditional electronic map making method has high cost and slow updating, so it is difficult to adapt to the current situation of frequent road network updating. Therefore, how to update the road network information quickly and effectively is studied, so as to provide people with the latest navigation. Path recommendation and other functions have a strong practical significance and broad application prospects. GPS data records the trajectory information of moving objects, and if these data are analyzed, it can provide auxiliary information for road network updating. Save the cost of making electronic maps. In addition, the reduced price of hardware devices with positioning capabilities and the rapid popularity of smartphones make it easy to collect road GPS data. Therefore, spatio-temporal data mining technology based on GPS trajectory has gradually become one of the research hotspots in the field of intelligent transportation system. As a commercial means of transportation, taxi has the advantages of long driving time, wide range of tracks, large cardinality of vehicles and low cost of data collection. The use of taxi GPS track data for road network updating and path selection has great advantages. Based on the open data set provided by Microsoft Asia Research Institute (MSRA) (Beijing taxi GPS data from April 2007 to October 2010), the following three aspects of work have been done: (1) GPS track preprocessing. Firstly, the data with errors is removed, and then the roads of the main roads in all trajectory data regions are obtained by using the rolling clustering algorithm. Finally, the accuracy of the road network is proved by comparing with Google Map. (2) A road network topology construction method based on taxi GPS trajectory is proposed. It is the premise of path navigation and road selection. In order to obtain up-to-date road and its connection relationship, this paper adopts the intersection recognition method based on taxi GPS track, using the road obtained in (1), taking into account the features of the number of road points at the intersection to extract the intersection. And further complete the road network topology construction, finally through the comparison with Google Map, Beijing shp map, proved the accuracy of the intersection; (3) A road selection method based on quad reactor is proposed. According to the path in (1) and the topological relation and the length of each road segment obtained in (2), combined with the historical GPS trajectory data, a Dijkstra method based on quad-heap priority queue is used to further realize the function of shortest and optimal path selection.
【學位授予單位】：湖南科技大學
【學位級別】：碩士
【學位授予年份】：2015
【分類號】：U495;TP311.13

【參考文獻】

相關期刊論文 前2條

1 孔令華,孔玲;淺談地圖的現(xiàn)狀與發(fā)展方向[J];勘察科學技術;2005年05期

2 李德仁,王樹良,李德毅,王新洲;論空間數據挖掘和知識發(fā)現(xiàn)的理論與方法[J];武漢大學學報(信息科學版);2002年03期

，

本文編號：2450463