摘要
在智能驾驶系统中,道路与障碍物的自动检测是最为关键的问题,也是安全行驶的基本保证。传统的检测方法与单目视觉检测都存在检测精度不高,鲁棒性不够等问题。一种基于立体视觉的自动识别算法,在建立三维道路模型的基础上,通过扩展卡尔曼滤波器的不断迭代与更新,能快速有效地识别出行驶过程中的车道边界,进而对车辆的位置与障碍物做出检测。
In the Driving Assistance System, roads and barriers automatic detection is not only the most critical issue, but also the basic guarantee of safety. The traditional detection and monocular detection have some disadvantages, such as accuracy and low robust. A lanes detection algorithm based on 3D can detect the lanes quickly and effectively during driving, so the vehicle's location and obstacles can also be detected robustly.
出处
《浙江国际海运职业技术学院学报》
2008年第1期4-8,共5页
Journal of Zhejiang International Maritime College
关键词
道路识别
道路跟踪
障碍物检测
扩展卡尔曼滤波
lanes detection
lanes tracking
obstacles detection
extended Kalman filter