Research On Road Recognition Technology Of Unmanned Vehicle Based On Binocular Vision

With the sharp increase in car ownership in the community,and this phenomenon has brought more and more serious traffic pressure and safety issues.Therefore,the study of driverless cars came into being,and become the world's top research area.Unmanned vehicles for environmental recognition through binocular vision,plays an important role in this area.In this paper,the key technologies were studied,which is belonged to binocular vision driverless cars road Recognition,Mainly related to the camera image distortion correction,image preprocessing,lane recognition,fuzzy control,identify obstacles and barriers to judge the distance.Both LabVIEW and NI Vision software as a programming platform in this paper,they complete the complex calculation and image processing.In the preparatory work for the completion of the acquisition of the original "image region of interest extraction,distortion correction,image graying,image filtering" and other research.Through the Hough transform and the original algorithm,this paper realizes the real-time transformation of lane lines with any degree of curvature to the five degree polynomial equation,this algorithm is not only suitable for the solid lane,also applies to the dotted lane.Recognition speed up to 60 frames per second,the recognition speed is extremely fast and the accuracy rate is extremely high.This equation is beneficial to the study of the extraction of the middle surface area of the fuzzy control and obstacle recognition.Fuzzy control and lane center curve equation,can reduce or eliminate the vehicle traveling direction and the deviation between the direction of the road in the shortest possible time to ensure that the vehicle traveling along the lane center requirements.Through this simulation,the control process is stable and reliable with good real-time performance.Realized the judgment of the dangerous obstacles and the calculation of the distance of dangerous obstacles.By using the known left and right lane curve equations,the road surface is extracted,and the interference of the objects on both sides of the road is ruled out,the obstacles are calculated by using the single-stress matrix,and the distance between the obstacles can be calculated by simplifying the projection matrix.The method of obstacle analysis is the innovation of this paper,the distance error is kept within 0.26%,the recognition result is accurate and reliable.In this paper,the contents of these studies were prepared corresponding program,through real-time analysis of the target image processing,the successful implementation of the initial objectives of this study,this paper proves the feasibility of road recognition algorithm.