Design And Implementation Of Vehicle Target Real-time Detection System Based On FPGA

In recent years,the number of cars has increased dramatically,which has brought great pressure to the traffic situation.The phenomenon of traffic congestion is becoming more and more serious.Traffic congestion brings many problems,such as low travel efficiency,environmental pollution,waste of oil resources and so on.Although the country has increased its investment in road construction,traffic congestion has not been fundamentally alleviated.In order to solve the problem of traffic congestion and improve the efficiency of traffic management,all countries in the world are building intelligent transportation systems.With the increase of surveillance cameras in cities,the video-based method to build intelligent transportation system is more and more widely used.In video based intelligent traffic system,accurate detection and recognition of vehicle targets through road monitoring videos is the basis of realizing other traffic management functions.In the past,vehicle target detection system was built on PC,which usually has the disadvantages of high cost,large volume and high power consumption.Moreover,the vehicle detection system is generally using the moving target detection algorithm,which can not accurately determine which is vehicle among the moving target,nor can it detect the stationary vehicle.This paper mainly studies the implementation of vehicle detection system on the heterogeneous computing platform of CPU+FPGA using Haar+Adaboost target detection framework.Haar+Adaboost can accurately identify the vehicle target in the video,no matter it is still or moving.FPGA can provide hardware acceleration performance for the detection algorithm and greatly improve the speed of the detection algorithm to meet the real-time requirements.By optimizing the target detection algorithm according to the calculation characteristics of FPGA and accelerating the heterogeneous computing platform,the whole system can meet the real-time detection of multiple targets and has certain usability.It provides a new solution for building video-based intelligent transportation system.