Research On Movement Patterns Mining And Visualization Analysis Of Moving Objects Based On Trajectory Data

The rapid development of location acquisition technology and various services enables researchers to acquire a large amount of movement data about multiple moving objects.Therefore,a large amount of movement data can be used to analyze various phenomena.At present,exploring the spatio-temporal movement patterns.behavioral characteristics and processes of moving objects is an important research hotspot.A key challenge in particular is to analyze and visualize a large movement dataset of' multiple moving objects to mine hidden spatio-temporal/spatial movement patterns,behavioral characteristics,and useful information and knowledge from high-dimensional and complex large data sets.This is driven by actual demands in a variety of applications,such as location-based services(e.g.navigation aids and ad placement),sociological studies(e.g.analysis of human behavior),traffic management(e.g.vehicle and pedestrian traffic control),law enforcement(e.g.video surveillance for criminal dctivities),physical analysis(e.g.the exploration of particle motion laws),weather forecasts(e.g.hurricane trajectory prediction and risk analysis),animal protection(e.g.tracking of endangered animal populations)and logistics management.Due to the diversity and complexity of moving objects,the rapid growth of data,and the demand from various real-world applications,the analysis and modeling of movement data of moving objects has become an important challenge for GIS discipline.In general,the mining and analysis of these movement data requires a combination of data extraction,analysis,statistical modeling,and quantitative and qualitative methods of geographic visualization.Visual exploration of movement data becomes a challenging task as a large amount of movement data becomes available.Visualizing these data tends to create confusing and incomprehensible images,so it requires complex filtering,simplification,or aggregation processes.Therefore.there need to be further exploration and research.In particular.researches in aggregate calculation,modeling,and visualization are needed to facilitate the development of methods and practices for mining useful intormation and knowledge from large and complex trajectory data sets.In summary,combined with the existing movement data analysis and visualization techniques,this research develops new trajectory data analysis methods and establishes different aggregate calculation models and analysis algorithms to carry out movement patterns mining and visual analysis research in two-dimensional space,three-dimensional space-time dimension and three-dimensional geospatial space.This work aims to make scientific contributions to the existing spatio-temporal/spatial analysis and modeling research of moving objects.The main research contents and results are as follows:(1)A series of theoretical basis and related knowledge of movement data analysis are systematically introduced.It involves many concepts such as:moving objects and movement trajectories,time geography framework.3D geospatial cube,quantitative analysis methods of movement data,qualitative visualization techniques of movement data,and 3D map algebra/multidimensional map algebra.(2)Based on the existing technology,the computational model and aggregation method of movement data in two-dimensional space are studied.For a small amount of movement data,a trajectory-based visual representation can be used directly.But for large and complex trajectory data sets,appropriate aggregation calculation methods are needed to extract useful information and knowledge about moving objects.The kernel aggregation techniques in two-dimensional space is quite mature.The existing kernel aggregation calculation methods of movement data are mainly divided into two categories:1)Spatial point pattern methods,it includes 2D point kernel density estimation and grid-based method.2)2D line segment kernel density estimation method(continuous model),it includes time-geographic kernel,Brown bridge kernel and directional kernel.(3)A new method of spatio-temporal aggregation for space-time trajectories in the framework of time geography is studied.It is mainly divided into three parts:1)A new space-time density estimation method(stacking space-time density)is proposed,which considers the spatial and temporal information to represent the spatio-temporal patterns and processes of moving objects in spatial and temporal dimensions.2)In the case study,the stacking space-time density algorithm was applied to the 30-day movement data of a transport truck in a mine in China to visually identify how the vehicle's movement behavior and patterns are spatially and temporally distributed.(4)Quantitative analysis and qualitative visualization of trajectory motion descriptors based on three-dimensional spatial density are studied.It Includes two parts,1)quantitative analysis and qualitative visualization methods:?In terms of quantitative analysis,it includes:a.Individual motion descriptors are used to describe the motion characteristics;b.Based on the mathematical operation of multiple motion descriptors,a new quantitative motion description indicator-4D time density is introduced into three-dimensional geographic space(x-y-z),and a novel method is provided to derive this quantitative indicator.?In terms of qualitative analysis,three-dimensional visualization of spatial density volumes based on volume visualization techniques(direct volume rendering,volume clipping plane,and isosurface)is implemented.2)The calculation principle and establishment process of the 4D time density algorithm are demonstrated by simulation data,and some basic visualization of the operation result is realized.(5)Three-dimensional visualization of density volume is realized by qualitative visualization techniques.Two case studies are involved:1)The spatial density based on the trajectory motion descriptor is applied to the flight trajectories of take-off and landing fhghts at Hong Kong International Airport and Macau International Airport.The experimental results show that the movement patterns and characteristics of a large number of flights can be described quantitatively and qualitatively.2)By comparing the traditional space-time density with the stacking space-time density,the visualization effect and computational efficiency of the two are compared.3)The evaluation and comparison of 2D kernel density estimation,stacking space-time density algorithm and 4D time density algorithm are realized.In this way,we can study,compare and visualize the differences of different models in describing the spatial and temporal distribution of movement patterns and characteristics of moving objects.