Design Of Vehicle Flow Detection System Based On Cortex

With the development of the economic society,private cars have been entering millions of households,becoming a necessary means of transportation for people to travel.Trucks,buses and so on,which are necessary for the development of industry and agriculture,make the roads in many cities more crowded.It is increasingly important for the traffic management department to obtain road traffic information and provide first-hand information for road management.It is also important for car owners to obtain road conditions,avoid congested roads,and reach destinations more safely and quickly.In this thesis,a Cortex multi-core processor X3399 is used as the detection platform.On the platform,OpenCV is transplanted,the background difference method is used as the vehicle target detection algorithm,and the CAMshift algorithm is used to track the target to achieve road traffic detection.The system uses a USB camera to collect video at the road site,then detects the traffic in real time and displays the detection results on the LCD screen in the form of a QT interface.In addition,the detection system can also directly detect the collected road vehicle video.In order to give full play to the advantages of the RK3399 multi-core processor and improve computing speed,OpenMP programming is used to achieve multi-core parallel acceleration for the system.This thesis mainly completes the following work:(1)Setting up the compilation environment to complete the system construction,including Uboot,linux kernel cutting and compilation,and file system construction.(2)Porting OpenCV and QT libraries,porting vehicle flow detection algorithm,background difference method combined with CAMshift algorithm to achieve vehicle flow detection.(3)OpenMP programming is used to achieve multi-core parallel acceleration.Image graying,background modeling using GMM models,background updating,and post-processing such as noise reduction and shadow suppression are optimized using OpenMP parallelization,Based on the traditional embedded single-core processing mode,the system operation speed becomes faster.(4)The test results show that the detection system basically reaches the design goal by testing the detection system on the road scene and analyzing the detecting effects and system performances.