基于优化ERFNet的智能车辆车道线精确检测算法

An algorithm for Accurate Lane Line Detection inIntelligent Vehicles Based on Optimized ERFNet

复杂场景中的车道线检测难度较高,为此,提出基于优化ERFNet的智能车辆车道线精确检测算法。利用弱瓶颈模块和非对称残差模块优化ERFNet模型,确定ERFNet模型结构参数,对车道线进行图像语义分割。通过图像切片和小线段提取车道线信息,基于拟合函数与残差函数实现智能车辆车道线的拟合检测。实验结果表明:不同道路场景测试样本下,所提算法的精确率和准确率较高,在不同天气条件下,检测时间较低。

Lane line detection in complex scenes is difficult,for this reason,an accurate lane line detection algorithm for intelligent vehicles based on optimized ERFNet is proposed.Optimize the ERFNet model using weak bottleneck modules and asymmetric residual modules,determine the structural parameters of the ERFNet model,and perform image semantic segmentation on lane lines.Extracting lane line information through image slicing and small line segment extraction,and implementing intelligent vehicle lane line fitting detection based on fitting functions and residual functions.The experimental results show that the proposed algorithm has higher accuracy and accuracy under different road scenario test samples and lower detection time under different weather conditions.