摘要
本文提出了一种基于机器视觉的车道检测与重建方法。采用链码算法检测分道线,用链码来记录车道线轮廓信息,再基于曲率模型的卡尔曼递推估计方法来估计当前的车道线位置,同时用来预测和检测下一时刻的车道线位置,并建立了车道数学模型。通过MATLAB仿真论证了该数学模型能准确地计算出真实车道的位置、弯度、形状等信息。但由于该数学模型计算较为繁琐,实时性不强,很难得到广泛应用。最后在该模型的基础上,通过计算相邻两段分道线的斜率差,简化算法,并对车道进行二维重建。实验结果表明,该方法能有效、快速地检测和重建车道,具有很好的可靠性和准确性。
A lane detection and rebuilding method based on machine vision is proposed. The lane-lines are detected by chain code algorithm, and the outline information of lane-lines is recorded using chain code. The present position of the lane-line is estimated by Kalman recursive estimation method based on curvature model; at the same time, the position of the lane-line in next moment is forecasted and detected, based on which the math model of the lane is built. MATLAB simulation results show that using this math model the information about position, curvature and shape of the real lane can be calculated accurately. But the math model is complicated in calculation, does not have good real time capability and can＇t be widely used. Finally, based on the math model, the calculation method is simplified through calculating the slope difference of two adjacent lane-lines, and the two dimensional rebuilding of the lane-lines is realized. Experiment results show that the proposed method can detect and rebuild lanes effectively and quickly, and has good reliability and accuracy.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2007年第7期1205-1210,共6页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(69674012)
科技型中小企业技术创新基金(04C26215110835)
重庆市信息产业发展资金(200311007)
重庆市自然科学基金(2006BA6016)资助项目
关键词
机器视觉
车道
链码检测
二维重建
machine vision
lane
chain code detection
two dimensional rebuilding
