Research On Vehicle Trajectory Matching Algorithm Based On GPS

Driven by the wave of Internet of Vehicles,location-based services(LBS)have become a popular item in the Internet of Vehicles products.The accuracy of vehicle positioning in these LBS products is an important factor affecting the functionality of these products.However,due to the limitations of the hardware,the sampling point must have error.At the same time,the map used by the map supplier also has certain errors,so it is necessary to collect the vehicle track point according to the provided map.Make certain corrections to achieve matching of track points and map,and improve the accuracy of vehicle positioning.This thesis mainly studies the online algorithm of trajectory matching for low-frequency GPS trajectory point.Focused on the following:(1)Android application to collect the GPS trajectory points of the vehicle is realized by JAVA,and the map is meshed by geo-hash algorithm,so as to use the error circle to retrieve the candidate segments of the track points in the map grid.(2)The features of trajectory points and road network are extracted based on the observation probability and the transition probability in the hidden Markov model.The observation probability is established by using the trajectory point to the road segment distance and the direction of the vehicle heading and the road segment direction.The transition probability is established by the similarity of the driving length and the path length and the similarity between the trajectory direction and the matching path direction.(3)Optimize the optimization algorithm for the online matching problem of low-frequency GPS points.The optimization algorithm is embodied in three aspects: Preprocess the trajectory point with speed 0,solving the problem of sampling hopping;Delay process according to the difference of joint probability,solving the mismatch of online matching;Seek the real path of the vehicle use error ellipse,and solve the roundabout path during low frequency sampling.(4)In order to verify the effectiveness of the improved path optimization algorithm,this thesis designs a vehicle trajectory matching system based on B/S architecture.The utility model comprises a data module formed by a GPS track point database and a map database,a trajectory matching module based on a Tomcat server,and a path display module based on a web interface.Finally,a comparison experiment is carried out for the preprocessing module,online processing module and path optimization module optimized by the optimization algorithm.The experimental results confirm that the optimized trajectory matching algorithm improves the accuracy of matching while ensuring real-time performance,and has a good effect on the online matching of low-frequency GPS trajectory.