The Design And Implementation Of A Lane Recognition System On Da Vinci Platform

In twenty-first Century,with the rapid development of information technology,intelligent technology has become more and more permeated with human life.Automatic driving system of intelligent vehicle has become a hotspot research with the development of artificial intelligence.In intelligent driving system based on intelligent vehicle vision navigation system,computer vision is the most critical technology for road recognition and tracking.In the automatic driving system of intelligent vehicle,the research of computer vision can be divided into monocular,binocular and multi eye vision navigation technology.Based on the research of intelligent vehicle automatic driving system and lane recognition technology,a multi-view lane recognition method and prototype system are realized combined with the road information of real road.The specific research results are as follows:(1)A lane detection and tracking algorithm based on the combination of road model and gray level feature is proposed.According the lane recognition method with high complexity and poor accuracy,a new road extraction method used Hough transforms is proposed.The proposed method implements accurate positioning and dynamic tracking of lane which combined with the feature of the real road and detected by gray feature and road line constraint model.(2)A multi-lane line fusion method based on multi-cameras visions proposed in this paper.Aiming at the single lane recognition method with weak robust,we propose coordinate transformation method by available lane lines extracted from multi-cameras,which fuse reference lane lines suitable for vehicle driving and prevent the system instability caused by part of cameras invalid.(3)A lane recognition system based on DaVinci platform is designed.In order to verify the effectiveness of the method proposed in this paper,the EMCV and embedded DSP technologies are used to transplant the lane recognition and detection algorithm on the DaVinci platform.Finally,to satisfy the real-time requirements,we optimize the lane recognition algorithm combined with the characteristics of DaVinci platform,which can achieve the highest efficiency on the premise of reliable extraction of the road ahead.In this paper,a variety of structured road videos have been collected under the urban environment and the system has been comprehensively tested.Experimental results show the stability,accuracy and efficiency of the system.