Research On Lane Line Detection System Based On FPGA

With the automotive industry,"the new four modernizations"continues to develop,autopilot technology,unmanned technology will gradually gained popularity in the smart car.Intelligent driving will be discussed at the national level at the National People’s Congress in2021,which will greatly promote the development of automation,intelligence and unmanned vehicles in the future.The lane line detection is an important research topic in the field of intelligent driving,the lane line detection technology can make the vehicle automatic obstacle avoidance,traffic information perception,lane departure warning and lane keep assist.In order to satisfy the requirement of the lane line detection accuracy,real-time,this paper put forward to improve the lane line detection Hough transform algorithm,and the FPGA core set the lane line detection system.The main research work is as follows:(1)The road image preprocessing.A series of image preprocessing algorithms are used to reduce the interference information and strengthen the lane information,aiming at the problems that noise and shade are not conducive to lane detection in the collected road images.Image preprocessing includes ROI extraction,grayscale,smoothing and edge detection.ROI extraction directly extracts the relevant information of lane features from road images,which can improve the detection speed.In order to keep the lane information fully,the weighted average method is used to process the grayscale.Smoothing processing to choose median filter,it is easy to appear in the road image noise filtering effect;Sobel operator is selected for edge detection,which can ensure real-time performance and effectively identify the characteristic information of lane lines,laying a foundation for the next step of lane line detection.(2)Optimal design of lane line detection algorithm.Aiming at the problems of the traditional Hough transform,which is time-consuming in computation and memory and difficult to set the quantization width of parameter space,this paper deduces the Hough transform algorithm which is suitable for the parallel computing structure in FPGA circuit,and makes it only have addition and displacement operation.Hough transform is then achieved by a multi-stage pipeline structure,the edge pixel corresponding to a compression parameter calculation in one clock cycle.In view of the fact that the traditional Hough transform will misidentify adjacent lane lines and non-lane lines,this paper adds constraints to the parameterθin the Hough transform so that the Hough transform can be calculated within the specified range,effectively removing the vertical and horizontal lane signs and others.Interference information.In further accordance with the principle of perspective,the method of comparisonθfilter parameters inside lane line,and left and right inner lane line to achieve the output parameters(ρ_L,θ_L)and(ρ_R,θ_R).(3)Build with FPGA as the core of the lane line detection system.In order to solve the problem of difficult algorithm transplantation and long time consuming,the lane line detection system is implemented by using top-down and pipelining design ideas.Complete the design with Verilog HDL language,mainly combining with the serial port to send module,TFT screen controller module,SDRAM controller module,image preprocessing module and the lane line detection module.Based on the Intel Cyclone IV E series FPGA as the development platform,make full use of the parallel processing and pipelining etc.,transmission,the road image preprocessing,the lane line detection and display.Finally the FPGA with MATLAB in the different stages of the lane line detection results and time efficiency comparison,verify the feasibility and efficiency of this system.