采用快速路面重建的车道和障碍物检测新方法  

New Method for Lane and Obstacle Detection Based on Fast Reconstruction of Road Surface

作　　者：郁梅[1] 蒋刚毅[1] 

机构地区：[1]宁波大学电路与系统研究所,宁波315211

出　　处：《仪器仪表学报》2002年第2期144-149,共6页Chinese Journal of Scientific Instrument

基　　金：浙江省青年人才基金 (RC0 1 0 57);浙江省自然科学基金 (No .60 1 0 1 7);宁波市科技攻关 (No .0 1J2 0 1 0 0 1 6)项目

摘　　要：对从二维图像平面中重建真实空间实际路面进行了研究 ,提出了实际路面快速重建算法 ,明显降低了其计算复杂度并保证重建的准确性。基于快速路面重建 ,本文提出了车道检测、障碍物检测方法并给出了相应的道路参数估计。实验系统仅需要一台摄像机获取路景图像 ,新方法能自适应地修正摄像机参数 。In this paper,the reconstruction of real road surface from an input image is studied,and a fast approach is proposed in order to reduce computational burden of the reconstruction without lose of robustness and reliability.Based on the fast reconstruction,a new method for lane and obstacle detection from a re organized image is proposed for driver assistance systems.In our experimental system,only one camera is adopted so as to reduce complexity and cost of system.Experimental results show that the proposed method can extract lane and detect obstacle effectively.

关 键 词：实际路面重建 透视投影 摄像机参数 车道检测 障碍物检测 智能交通系统 

分 类 号：U491.2[交通运输工程—交通运输规划与管理] U495[交通运输工程—道路与铁道工程]