Research On Traffic Signal Recognition Method Of Micro-vehicles At Crossroads Based On Vision

As a hotspot of technology development in the current scientific and technological community,the main research contents of autonomous driving technology include environmental perception,decision-making and planning,control and execution.The perception of three traffic signals for traffic lights,road signs and parking lines is the key to autonomous driving.In this paper,the micro-traffic platform and micro-vehicle are built according to the micro-scale ratio of 1:10.In view of the previous study,only a single traffic signal is considered to control the shortage of micro-vehicle.The method of image-based image processing and analysis is adopted to realize traffic lights and roads.The accurate identification of the indicator and the parking line completed the automatic driving experiment of the micro-vehicle through the intersection.The specific contents of this thesis are as shown:(1)Based on the 1:10 scale,the micro-transport platform and micro-vehicle were designed and built.The micro-transportation platform contains a variety of traffic environment elements,which restores the real traffic scene;the micro-vehicle adopts a modular design,and can independently realize driving,deceleration parking,steering and other functions according to the command;(2)Aiming at the problem of detection and identification of traffic lights,a system framework for detection,identification and tracking of traffic lights has been established.The framework is easy to implement and has high detection accuracy.In the detection phase,the image to be recognized is color-divided in the HSI color space and the segmentation result is filtered to obtain a candidate region including the traffic light;in the recognition phase,the circularity detection is performed on the candidate region,and the traffic light region of the traffic light is accurately positioned,and The area is subjected to hue histogram statistics,and the traffic light color recognition is realized according to the hue information;in the tracking phase,the traffic tracking based on CamShift is used to track the traffic light area in real time;(3)Aiming at the problem of identification of road marking indication,the area containing the indication information is preprocessed by inverse perspective transformation,and the image is segmented by the method of obtaining the optimal threshold based on the average gray level.Finally,the improved K-nearest neighbor algorithm is used to identify the road markings;(4)As for detection of parking line,firstly filter the irrelevant information areas such as the zebra crossing at the crossroads,then use the LSD(Line Segment Detection)line detection algorithm to detect the parking line;(5)In this thesis,the traffic signals of traffic lights,road signs and parking lines are comprehensively studied.The fuzzy control algorithm and incremental PID(Pelvic Inflammatory Disease)control algorithm are used to adjust the steering and speed of the micro-vehicle respectively.The traffic signal recognition of the micro-vehicle at the intersection is completed.experiment.The experimental results show that the microvehicle can accurately identify the traffic lights,road signs and parking lines in the microtraffic environment,and complete the control of the speed and steering.