Design And Implementation Of Traffic Flow Detection System Based On Distributed Intelligent Video Analysis

With the increasing traffic flow of urban roads,how to resolve traffic congestion with high efficiency and low cost is an important research content in the field of intelligent transportation.Traffic flow statistics based on video analysis technology provides a reference for traffic management department in congestion dispersal.Due to some factors such as vehicle density and road conditions,which lead to different traffic flow statistics in different traffic density scenarios,this thesis proposes a distributed traffic flow detection method based on split-video and frame difference method for low-density traffic scenarios.Aiming at high-density traffic scenarios,a traffic flow detection method based on video frame segment distributed processing and background difference method is proposed.Finally,this thesis achieves an efficient traffic flow detection system.The main research work of this thesis is as follows:(1)This thesis researches a distributed traffic flow detection method on low-density.When the existing detection method performs traffic flow statistics under a scene with low traffic density,there is usually a problem that the detection speed is slow.In this thesis,the complete video is split into a number of short videos in advance,and a video reading program on Spark is designed.Then,a Frame-Difference method is used to realize distributed statistics of traffic flow in videos,which increases speed of traffic flow statistics.(2)This thesis researches a distributed traffic flow detection method on high-density.When we perform traffic flow statistics under a scene with high traffic density,traditional detection methods have problems of missing detection and inefficient statistics.By researching the distributed reading method of the video frame segment and the Spark data tilting solution,this thesis proposes a video frame segment reading scheme and task average distribution method,and realizes distributed traffic flow statistics on video frame segments based on the Background-Difference method,which improves the detection speed while ensuring high accuracy.(3)This thesis designs and implements a traffic flow detection system on low and high density.According to system business requirements,functional modules are analyzed and designed to realize distributed traffic flow statistics and display the statistical results under different traffic density scenarios.The system is constructed by Spring,SpringMVC,Hibernate,etc.Through the function test and performance test of the system,it is verified that the system has faster detection speed and higher accuracy under different scenarios.The research results of this thesis can be applied in real scenes and contribute to the development of traffic flow detection research.