Design And Implementation Of Distributed Trajectory Generation System For Target Personnel

With the continuous expansion of the city scale,traditional manual methods to monitor and investigate target personnel are costly and inefficient.However,with the development of image recognition technology,the facial features of the target personnel can be extracted from the face pictures collected by the surveillance camera.Depending on these features,the target personnel's trajectory information can be mined,providing an important reference for public security organs.However,surveillance data has the characteristics of large data volume,low data density and obvious spatiotemporal properties.Therefore,how to efficiently store and analyze massive surveillance data is a topic of great research value.Based on the above background,this thesis designs and implements a distributed trajectory generation system for target personnel,which can complete the trajectory generation tasks of target personnel by performing distributed storage and calculation of surveillance data from multiple sources,improve the efficiency of trajectory generation tasks from data access and trajectory calculation,and quickly complete the trajectory tasks such as generating trajectory information,generating foothold information,finding people who appear with the target personnel at the same time and finding people who accompany target personnel within a few seconds.This thesis mainly completes the following tasks:1.Design and implement a distributed trajectory generation system with a masterslave architecture.The master node is responsible for task scheduling and resource management,and the slave nodes are responsible for completing specific storage and trajectory generation tasks.2.Design and implement a distributed storage engine optimized for the spatiotemporal properties of surveillance data.The engine uses a multi-memory cache strategy and two-dimensional spatiotemporal indexing technology to achieve efficient concurrent access to multiple surveillance data.3.Based on the distributed data storage engine and the spatiotemporal properties of the surveillance data,a distributed trajectory generation engine is designed and implemented.The engine contains a distributed clustering operator and a distributed association rule mining operator.The implementation of the distributed clustering operator is based on the K-Means algorithms and DBSCAN algorithms.It uses data sharding,multi-core concurrency,search and merge,and data proximity calculation to optimize the spatiotemporal properties of the surveillance data and improve the clustering efficiency.The implementation of the distributed association rule mining operator is based on the Fp-Growth algorithm,which uses distributed support counting and distributed frequent itemset mining to improve the efficiency of partner analysis.4.This thesis conducts detailed functional and performance testing of the distributed trajectory generation system for target personnel,and analyzes the test results in detail.