摘要
车道检测算法的研究是室外移动机器人基于道路区域或边界信息自动导航的首要环节。根据实际应用需要,识别算法应能满足鲁棒、实时的要求。提出的基于全局模板的快速车道检测算法,根据驾驶员视觉处理经验,总结出道路边界识别的先验知识,在所选取的二次曲线道路形状模型基础上,推导出图像上车道的总体模型,利用道路图像的灰度特征,结合优化算法改变模板参数,实现对车道的跟踪。对实际路面图像的试验结果表明,该算法在多种环境条件下,都能很好地识别车道,具有很强的鲁棒性。同时,与传统的跟踪算法相比,该算法降低了计算损耗,更好地满足了实时性的要求。
It＇s principal to detect lane robustly and rapidly for outdoor robots, based on the information of road edge or road region. Lane detection method with deformable temple described here first presented pre-knowledge based on human visual experience. With the selected road shape model, lane model in image plant was introduced. Using intensity feature of lane image, an optimization algorithm was established to maximize likelihood function evaluating how well the image gradient data on an assumed lane marking supports a given set of template parameters. In the concluding, the result of the real road tracking experimentation was analyzed, which proved the algorithm could robustly detect lane markings in situations with rain, shadows or broken. Compared with traditional lane detector based on image intensity gradient, the method decreased the cost of computation, and further increased real-time performance.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2007年第21期5063-5066,共4页
Journal of System Simulation
基金
江苏省高技术研究计划(BG2005014)
关键词
移动机器人
车道检测
变形模板
函数优化
mobile robots
lane detection
deformable template
function optimization
