车道线信息的全面理解及偏离预警算法

为全面理解车道线信息,提出了一种车道线检测分类跟踪及偏离预警算法。首先利用动态感兴趣区域约束Canny算子的检测范围,基于扩展的Otsu算法改进Canny算子的阈值设定方式,并通过Hough变换进行车道线边缘拟合;然后依据车道线的颜色及线型特征进行分类,同时借助Kalman滤波器实时跟踪车道线,对检测失效区域采用Kalman滤波器的预测值进行替换;最后设定有效的偏离预警策略,确保行驶的安全性。实验结果表明,算法能全面地理解车道线信息并进行跟踪,同时具备对危险行驶状态下的车辆进行预警的能力。

Comprehensive understanding and departure warning algorithm of lane line information

To fully understand lane line information,a lane line detection classification tracking and deviation warning algorithm is proposed.Firstly,the detection range of Canny operator is constrained effectively by dynamic region of interest algorithm.And the threshold setting mode of Canny operator is improved by extended Otsu algorithm.Hough transform is used to fit the lane line edge.Then the lane line is classified according to color and line-type characteristics.Meanwhile,the lane line is tracked in real time through Kalman filter.The predicted value of the Kalman filter is used to replace the failed lane line.Finally,an effective departure warning strategy is established to guarantee driving safety.The experimental results show that the algorithm can fully understand lane line information and track it,which has the ability to warn vehicles in dangerous driving conditions.