摘要
基于机器视觉的弯道图像能提供车辆行驶道路环境的丰富信息,从建立弯道模型、提取车道线像素点以及拟合车道模型等步骤分析了传统基于道路模型的弯道检测方法,针对传统方法很难适用于多种不同形状弯道的特点,提出一种基于特征点提取的弯道检测新方法;介绍了弯道检测在车道偏离预警、弯道限速以及弯道防碰撞预警等领域的应用情况;最后提出弯道检测应该建立三维车道线模型、注重发展多传感器融合技术,提高其适用性和鲁棒性。
Curved road images based on machine vision can provide a lot of information about driving environment.The traditional curved road detection methods are analyzed according to the detection steps.The detection steps includes establishment of curve model,extracting lane pixels and fitting of lane model.The traditional curved road detection methods are not appropriate for different types of curved roads.A new curved road detection method based on feature point extraction is proposed.Then,the applications of warning systems based on the curved road detection are presented,including lane departure warning system,speed warning system,front collision warning system for curved highway and so on.Finally,future research on curved road detection is advised to establish a three-dimensional lane model and to focus on the development of multi-sensor fusion technology to improve its applicability and robustness.
出处
《交通信息与安全》
2012年第3期141-146,共6页
Journal of Transport Information and Safety
基金
国家自然科学基金项目(批准号:61104165)资助
关键词
机器视觉
道路模型
弯道检测
预警系统
machine vision
road model
curved road detection
warning system
