Discovery And Analysis Of Chasing Patterns In Trajectory Data

Global positioning system,wireless sensor network and social media applications contain a lot of trajectory data.These trajectory data contain abundant spatial and temporal information of moving objects such as people,vehicles,animals and clouds.Through calculation and analysis,we can find the behavior patterns of moving objects,such as chasing,leadership,avoidance and other types of patterns.Among them,the discovery and analysis of the chasing pattern can provide effective services for intelligent urban traffic management,military reconnaissance,criminal investigation,environmental monitoring and other application fields.The existing chasing pattern matching methods mainly have the following problems:1)without considering the road traffic network(i.e.road network),and only comparing the local similarity of two trajectories,so that some non-existent chasing patterns are"found"in complex road network scenarios;2)without differentiating the trajectory data corresponding to the local behavior of moving objects such as staying and changing speed,resulting in some trajectory data.The chasing behavior with ulterior motives is not found effectively;3)In the chasing pattern matching,the method of representing line segments is used to calculate the distance,so it is difficult to find the chasing behavior with large angle direction change.Other pattern discovery methods related to trajectory are used to track patterns.There are many problems in the fashion of pattern discovery,such as large computational load,low efficiency and inaccuracy.In order to solve the above problems,this paper specially explores the discovery and analysis method of chasing pattern based on trajectory data,systematically studies the representation of trajectory,the calculation of trajectory similarity,the classification and characteristics of trajectory pattern,the specific meaning of chasing pattern,designs the model of chasing pattern discovery and analysis,and introduces the technology of map matching and stay point detection in the model to improve the chasing pattern.The accuracy,recall rate and efficiency of the discovery are analyzed,and a series of algorithms are designed.Experiments and theoretical analysis verify the effectiveness of the model and algorithm.The main contributions of this paper are as follows:(1)A general trajectory chasing pattern discovery and analysis model(DACPINT model)is designed,which is suitable for the discovery and analysis of multiple traj ectory chasing patterns.(2)In the DACPINT model,the application of map matching technology in the discovery of chasing patterns is proposed,the road network trajectory is established,and the chasing pattern discovery algorithm based on the road network trajectory is designed.The algorithm is suitable for vehicle trajectories and can eliminate pseudo-similar trajectories in different paths.Because of the influence of urban buildings and road network,and the speed is fast,there are many possibilities for the vehicle's path between two sampling points.Therefore,it is particularly important to obtain the most likely path for the vehicle in the chasing pattern discovery.The application of high-precision map matching technology can improve the accuracy of the chasing pattern discovery.(3)The stay point detection technology is studied,and the optimization algorithm is designed.In the DACPINT model,the stay point detection technology is used to assist the discovery of the chasing pattern,and the chasing pattern discovery algorithm based on the stay point optimization algorithm is designed.The chasing pattern discovery algorithm can be used not only in the trajectory with road network reference,but also in the trajectory without road network reference.It can improve the recall rate of chasing pattern discovery,and further improve the accuracy and efficiency.In the process of moving,the object will stop when it is blocked or passes through the points,regions and events of interest,thus affecting the behavior pattern of chasing.Therefore,by optimizing the stay point detection technology,on the one hand,the algorithm can eliminate the trajectory without obvious chasing intention;on the other hand,it can find the potential chasing pattern in the trajectory,and can find more complete chasing pattern,not only the local fragmentation chasing pattern.In addition,this paper also optimizes the trajectory similarity calculation method in the chasing pattern discovery algorithm based on the stay point optimization algorithm,so that the chasing pattern can be effectively found when the trajectory shape changes greatly.(4)Through a series of experiments,the validity and performance of the proposed DACPINT model and the above chasing pattern discovery algorithm are verified.Aiming at the trajectory data corresponding to the discovered chasing pattern,the chasing confidence under time and space constraints were analyzed,and the possible intention of the chasing behavior was evaluated.