Research On Lane Detection Of Intelligent Micro Vehicle Based On Monocular Vision

Facing the increasingly serious traffic problems,such as traffic jam,serious environmental pollution and frequent traffic accidents,the development of intelligent vehicles and the demand for technology upgrading become particularly urgent.Intelligent vehicle have the same function of artificially driving as traditional vehicles.Through a variety of controlling hardware,executing hardware and other sensors and son on,with the excellent sensing algorithm,vehicles can perceive the traffic environment intelligently,and can deliver and feedback the perceived information to itself in real time,also can analyze and judge the vehicle driving status.Thus,various decisions in the process of driving can be done intelligently under the unmanned intervention.However,lane line is an important basis for intelligent vehicle to perceive the surrounding environment and make driving decisions.Lane detection is one of the key technologies of intelligent vehicle.Therefore,in the background of intelligent vehicle and lane detectio n technology,this thesis researches on lane detection of intelligent micro vehicle based on monocular vision.The main contents of this thesis are as following:Chapter ? introduces the back ground and meaning of this thesis and analyzes the present state of relative researches.Based on that,the overall structure of this thesis is introduced.Chapter ? introduces the composition and working principle of the intelligent micro vehicle hardware system,explains the design plan of intelligent micro vehicle,and designs the communication network,then develops the upper computer software through Visual Studio2010.Chapter ? researches and uses monocular vision technique,and designs the the lane feature extracting scheme.Then image preprocessing was carried out by using grayscale,Sobel operator edge extraction and positive and negative extreme point extraction method.Combining with the inverse perspective transformation results,this thesis focuses on the innovation lane extraction method based on lane width and continuity of the priori information.Thus,it can significantly reduce noise points of the lane detection.Chapter ? compares and analyzes the shortcomings of the straight line detection method,analyzes and proposes an improved algorithm based on RANSAC to improve the adaptability and efficiency of the model fitting.The lane detection algorithm is tested under different conditions in the simulation environment and urban road.The experimental results show that the algorithm is effective and robust.Chapter ? summarizes the results of this thesis and discusses the future work of the research.