Research And Implementation Of FPGA Acceleration For Target Vehicle Detection Algorithm

With the rapid growth of car ownership,the safety of vehicles while driving on the road becomes more and more important,so people’s attention to intelligent driving is also increasing.A safe driving process requires the driver to quickly identify and react to the driving environment to avoid a car accident.The Advanced Assisted Driving System(ADAS)is an integrated control system with active safety that helps the driver to make environmental assessments of road conditions,pedestrians,vehicles,etc.,and then alert or help.This system uses sensors such as cameras to collect data related to the driving environment,and provides effective driving assistance information for the driver by identifying and detecting the vehicle in front,thereby improving safety and comfort during driving.Based on the in-depth study of the vehicle detection system and algorithm,based on the software and hardware collaborative design,LBP feature extraction and AdaBoost classifier are selected as the target vehicle detection algorithm.The development tool SDSoC from Xilinx is used on the Zynq-7020 development board.Achieve front vehicle detection.The main research contents and results of this paper include:(1)The basic theory and method of vehicle detection are expounded,including several common vehicle feature extraction algorithms and several commonly used classification recognition algorithms.The comparison feature extraction algorithm and classifier select the vehicle detection algorithm suitable for this system.(2)Research on key technologies for FPGA acceleration algorithms.After analyzing the hardware acceleration devices of GPU,ASIC and FPGA,select the acceleration device suitable for the system,carry out the software and hardware collaborative design,and then refine the design hardware,architecture and process.(3)Complete the implementation of the vehicle detection acceleration system.The video acquisition,video buffering,image pre-processing and display module of the programmable logic part are realized in the development environment,and the feature extraction and classifier algorithm and initialization program module of the processing system part are realized.(4)System experiments and results analysis were completed.On the basis of successfully setting up the running platform,the system is running,and the advantages of using FPGA acceleration are clarified by analyzing and comparing the quantitative data of various aspects.