Design And Implementation Of Video Surveillance System Based On Cloud Platform

With the rapid development of the Internet,video surveillance technology has been widely used.Today,video surveillance equipment is deployed in all corners of daily life,providing people with great convenience and security.The management of a large number of video surveillance equipment has always been a critical issue in the field of video surveillance.Video surveillance systems based on traditional stand-alone architectures are difficult to bear the increasing burden of computing and cannot effectively cope with massive video request services.This thesis studies the construction of a cloud platform video surveillance system based on Docker container technology to solve the increasingly large user scale and massive video requests.At the same time,it develops a supporting video surveillance client application.The main work of the thesis includes:Firstly,the docker cluster is built based on the docker container technology,and the developed video forwarding service,client management system,login verification service and video cache management services are deployed on the cluster..The video forwarding service is developed based on the open source framework Easy Darwin and supports video streaming and forwarding functions,as well as distributed deployment.The client management system is developed based on JAVA and supports effective management of clients and a large number of video surveillance equipment.Secondly,a video surveillance client application has been developed based on the Android system platform.The application supports hardware decoding of streaming media.When playing video,it can reduce the occupancy rate of the mobile phone processor and improve decoding efficiency.Finally,the paper studies the task scheduling algorithm of Docker cluster.The Swarm cluster management tool provided by Docker has its own Spread task scheduling strategy,but the algorithm only considers the utilization of CPU and memory resources,and the algorithm is easy to cause node load imbalance.The dynamic load balancing algorithm introduces the utilization of network resources and secondary adjustment factors.On the basis of this algorithm,this paper uses a prediction algorithm to predict the change trend of resource utilization rate,and uses a nonlinear quadratic adjustment factor to dynamically adjust the weight parameters.Simulation experiment results in the LAN environment show that the improved algorithm can improve the overall resource utilization of the cluster.The test results show that the video surveillance cloud platform designed in this paper has achieved the functions required by the design.