The Application Of Machine Vision Technology In The Intelligent Control For Vehicle

With the rapid development of economy and technology in our country,more and more intelligent vehicles are appearing in people’s lives.The application of machine vision theory in intelligent vehicles is increasingly mature,which has become an opportunity for its application in intelligent vehicles.This paper focuses on the related algorithms around the application of machine vision technology in intelligent vehicles,and designs an experimental platform for it.The specific work is as follows:Firstly,aiming at the problem of lane detection during vehicle driving,an optimal lane detection algorithm is proposed,based on mathematical morphological features.First of all,he images taken by the vehicle camera are grayed,filtered,Canny edge detection and PPHT to extract lane line information.Then,the optimal left and right lane lines are selected according to the mathematical morphological characteristics of the lane lines on the image.Lastly,compared with the lane line detection program composed of Open CV related native functions,it achieves better processing effect.Secondly,aiming at the problem of pedestrian detection before the car during vehicle driving,a pedestrian detection program is developed based on HOG + SVM.INRIA Person Dataset is used for training and Caltech Pedestrian Detection Benchmark for testing.From the test results,it basically meets the requirements of early warning for pedestrians in front of the vehicle.Thirdly,it studies the control of vehicle by analyzing the vehicle’s visual model,two-degree-of-freedom dynamic model and road(lane line)model.The lateral and longitudinal controllers of the vehicle are designed based on fuzzy control and digital PID respectively and simulated in MATLAB software.By comparing the simulation results,the controller can meet the basic requirements of the tracking task.Finally,The experimental platform was implemented and the verification experiment was carried out.First of all,a hardware organization structure in which video communication and vehicle master control are completely separated is proposed,and the hardware system of the experimental platform is built.Then,written the software system of the experimental platform(it mainly includes vehicle main control program based on STM32F103RCT6,video network communication program based on Linux kernel and upper computer control software based on C#).Lastly,experiments on this experimental platform show that the effectiveness,compatibility and practicability of the lane detection method and controller in intelligent vehicle applications also indicate that the experimental platform can stably and effectively complete machine vision technology in vehicles.Relevant verification tests in intelligent control.