The Research Of Image Processing Method About Lane Departure Warning Based On DSP

Lane departure warning system based on DSP is one of the key technologies of activesafety of car, which is also the hotspot in the field of automatic control at home and abroad.Based on the core processor TMS320DM6437of TI, this paper analysed and researched animage processing method of lane departure warning system on the hardware developmentplatform of ICETEK-DM6437-B eval board made RuiTai company, combined with softwareintegrated development environment CCS3.3.After the introduction of relevant background and development status of area for researchon lane departure warning system at home and abroad, this paper elaborated an imageprocessing method for lane departure warning system based on DSP that can achieve theextraction of highway structural road lane line on the premise of the accuracy and real-time.Its specific contents are as follows:Firstly, the paper had preprocessed the road image, that reduced the amount of data of theimage processing stage by grey level transformation and effective regional segmentation;Median filter is selected as the de-noising method through comparative analysis of differentfiltering scopes and treatment effect; Enhanced road image contrast by histogram equalization.On the stage of detection and extraction of lane line. The paper detected the centerline seedpoint of lane line by horizontal and vertical scanning after binary segmentation on road imagewith selecting reasonable threshold values, and successfully fitted the lane line based on thecreation of lane model, combined with the least squares method. Then, the article haddesigned the lane departure warning decision-making by making the bisector angle of thevehicle centerline and lane interior angles as the yaw angle of system, determined that thevehicle would be about to deviate from the lane line when the angle exceeded the thresholdrange, and reflected the result of the judgment to the display by the lane line marked withdifferent colors. Finally, road image processing based on DSP had been achieved. It designedsoftware scheme and wrote algorithm program; Adjusted and identified threshold valuesneeded in the image processing by real vehicle experiments, then analyzed and verified theeffect of lane line fitting and warning. Experimental results showed that lane recognition algorithm based on DSP in this paper had high accuracy and commodity in the field of safetyintelligent driving.