Visual Perception And Recognition Of Vehicular Access Under Complex Conditions

The road identification under the complex conditions is the hot topic of current intelligent vehicle-assisted safety driving sub-proposition,is also a common problem in the field of transportation.In this paper,a variety of imaging methods are introduced into the traditional lane mark detection technology for the road environment under external complicated conditions,and the more real-time,more adaptable and more robust lane recognition algorithm is explored to build new road visual perception and recognition system.Firstly,this paper introduces the image acquisition system needed for research,and lists the main fundamental theories of image processing.And then the road conditions under varieties of complex road traffic are analyzed,according to the details of image characteristics,the corresponding solutions are explored.Then,by studying the basic characteristics and design specification of the road structure and combining with the visual imaging characteristics of the road plane linear structure,the roadway line straight-curve mathematical model was constructed.For the near vision field straightening,an improved Hough transform linear detection algorithm to improve the accuracy of the lane mark detection and recognition.Compare with the traditional Hough transform traversal full graph detection,it also save the time.In this paper,a new method of turning lane recognition is proposed for the far-field corridor lane.The method of regional growth is used to find the end point of the curve lane.The tangential characteristics of the straight lane and the curve lane.Combined,the mathematical model about turn lane is solved,and then the turn lane is extracted.The extracted road lane contains detailed real-time traffic conditions and road information,which are explored and used,a piece of very valuable vehicle driving warning information is presented.The main achievements and innovations of this paper are as follows:(1)A new vehicle deviation detection algorithm is proposed to establish the mathematical model of vehicle deviation to realize the judgment of the current vehicle deviation.(2)A new method for detecting the slope change is designed,The difference between the image slop of unchanged gradient and that of the current 30 consecutive images is compared to achieve the judge of the slope for road ahead in a very short time.Through the simulation and experimental verification,the algorithm designed in this paper can detect and identify the road lane line in complex environment,and can accurately assess the deviation of the direction of the vehicle and the slope of the road,which is very important for the research of automobile safety assistance driving.