Study On Urban Road Network Information Extraction Method Based On Floating Vehicle Trajectory Data

As the most important part of modern transportation system,urban road network is closely related to residents’ life and social development.With the development of China’s economy and the improvement of infrastructure,the urban road network structure is also changing rapidly.How to make the digital map quickly and accurately obtain the latest road network status and update it in time has become particularly important.There are two traditional road network construction methods.One is to use special vehicles equipped with high-precision global positioning system equipment for field survey.The other is extracted from remote sensing satellite images.The road network extracted by these two methods has high accuracy.The disadvantage is that it needs professional team operation,which is expensive and time-consuming.It is not suitable for the acquisition and updating of large-scale road network.As a common means of transportation in residents’ daily life,taxi has attracted the attention of many scholars in recent years because of its advantages of low data acquisition cost,strong real-time and wide coverage.It has also made great contributions to the construction of digital road.This paper focuses on the low-frequency and low-precision taxi trajectory data,and designs a road network construction scheme.The specific research contents are as follows:(1)Trajectory data preprocessingAiming at the problems of data redundancy and too many noise points in the original trajectory data,a de redundancy algorithm based on velocity threshold and a heuristic filtering algorithm based on density are proposed to preprocess the trajectory data;For the problem of missing individual field values,the algorithm assisted method is used to roughly estimate.(2)Road intersection extraction based on density clustering algorithmOn the basis of data preprocessing and according to the trajectory characteristics of taxis,a step-by-step road intersection feature point recognition method is proposed,and the centroid of Road intersection feature points is extracted by clustering algorithm,which is the final road intersection.(3)Road skeleton extraction based on KD tree and multivariate adaptive regression splineFor the extraction of road skeleton,this paper proposes two solutions: the road skeleton extraction scheme based on KD tree structure and the road skeleton extraction scheme based on multivariate adaptive regression spline(MARS)method.The extraction effects of the two methods are compared horizontally,and the best is selected as the final scheme.(4)Establishment of road network topologyBased on the extraction of road intersections and road skeleton lines,the topological relationship among road trace points,road intersections and road vector edges is analyzed,the topological structure of road network is constructed,and the vector road network structure with topological properties is generated.