Research On Road Marking Damage Detection And Repair Technology Based On Machine Vision

As an important guide to ensure the safe driving of vehicles,road markings often convey accurate traffic information to drivers or pedestrians through symbols and words.However,after being used for a period of time,it will produce damage and aging,which will cause hidden dangers to traffic safety.The maintenance of road markings mainly depends on manual on-site visual inspection and manual repair.In order to improve the maintenance efficiency,this paper designs a road marking damage detection and repair system based on machine vision,studies the realization principle of the vision system,and puts forward the method of installing a camera at the bottom of the marking car to collect image data,and forms the overall scheme of the whole system,mainly including:Video image acquisition and image preprocessing,road marking image damage detection,damage repair.Firstly,the ov7725 camera module,SDRAM storage module and VGA display module are designed respectively by building the image acquisition system.The simulation verification is realized with Modelsim software,and the camera is calibrated with MATLAB software to obtain the key parameters of the camera.Secondly,the image preprocessing method is studied.The road marking image is transformed into gray image,and the histogram equalization method is used to distinguish the road area from the marking area in the image,so as to enhance the image contrast.Then through the comparative analysis of the three filtering algorithms through simulation experiments,taking the median filtering algorithm is the first choice for noise reduction,the image which is closer to the real environment can be obtained.The hardware implementation process of the related algorithms is introduced.Then,Otsu algorithm is used to realize image segmentation.The preprocessed road marking image is transformed into binary image,and the small defects of binary image are removed by using image morphology knowledge.The FPGA implementation methods of two related algorithms are introduced.Taking the largest circumscribed rectangle as the original marking area,the damage rate is calculated according to the ratio of the number of damaged pixels to the number of pixels in the intact marking area,and the damage is divided into different processing states.In order to verify the reliability of the whole algorithm,the Matlab GUI interface is designed.The experimental results are consistent with the reality after repeated verification of multiple images of different road sections.Finally,considering that the repair of small damage has no practical significance,the elimination criteria are designed to eliminate small damage.In order to realize the repair of irregular damage,a bidirectional controllable repair device model is established,and the opening size of the discharge device is flexibly controlled by PWM control method.The implementation process of repair is described in detail.Through physical debugging,the design meets the requirements.Finally,the visual platform system is built for testing.The experimental results show that the system can detect the damage of road markings and repair them in real time.The damage rate and image processing results of road markings are the same as those of MATLAB simulation.When the speed of the system is 0.5m/s,the repair effect is the best.