摘要
车道线识别方法的突破对于无人驾驶汽车的研究具有重大意义,文章分析了基于深度学习的两种车道线检测算法,利用了视觉传感器图像处理技术对车道线进行了识别,研究了语义分割的算法。视觉图像处理则分别用目标检测算法和语义分割算法对车道线进行了识别,对两种算法做出了分析比较。研究结果表明:基于深度学习的语义分割网络具有更好的车道线识别能力;能得到较为可靠的车道线边界信息。文章的车道线识别方法可以为无人驾驶汽车提供较为可靠的车道线数据信息。
In this paper,two lane lane detection algorithms based on deep learning are described,and lane lane is identified by visual sensor image processing technology.In visual image processing,lane lines are identified by target detection algorithm and se-mantic segmentation algorithm,and the two algorithms are analyzed and compared.The results show that the semantic segmentation network based on deep learning has better lane lane recognition ability.Reliable lane boundary information can be obtained.The br-eakthrough of lane line identification is of great significance to the research and development of driverless cars.The lane line identi-fication method in this paper can provide reliable lane line data information for driverless cars.
作者
翟煕照
韩同群
杨正才
Zhai Xizhao;Han Tongqun;Yang Zhengcai(Hubei Institute of Automobile Engineering,Hubei Shiyan 442002)
出处
《汽车实用技术》
2021年第6期29-34,共6页
Automobile Applied Technology
基金
“汽车动力传动与电子控制湖北省重点实验室”汽车零部件技术湖北省协同创新项目(2015XTZX0425)。
关键词
图像处理
无人驾驶
深度学习
语义分割
Image processing
Unmanned
Deep learning
Semantic segmentation