【作者】 刘化胜；

【导师】 王友仁；

【作者基本信息】 南京航空航天大学 ， 电力电子与电力传动， 2004， 硕士

【摘要】 视觉导航是智能车辆导航的一项关键技术，本文针对智能车辆视觉导航系统研究了基于图像检测的道路检测算法。 论文介绍了智能车辆视觉导航系统研究背景、意义和现状，分析了常用道路图像预处理算法，提出了一种新的道路检测算法。文中算法采用基于线段的区域增长法将道路边缘图像分割成道路区域和非道路区域，对于多车道的检测，根据道路标志的边缘，恢复道路标志并根据其特征进行定位，最后采用数据拟合的方法找出道路轨迹线。按照本文算法在复杂路况下可以准确、快速估算出车道的延伸方向，从而可以实现车辆的防偏预报。 利用采集的200幅实际道路图像对本文提出的算法进行验证，正确识别率达81.2％，仿真结果表明本文提出的道路检测算法是可行和有效的。更多还原

【Abstract】 Vision navigation is the key technique of intelligent vehicle guidance, this paper researches the algorithm of lane detection for intelligent vehicles vision navigation system based image process.In the paper, research background, meaning and current status of vision navigation is firstly introduced, and currently algorithm of image process is analyzed; secondly, new algorithm for lane detection is presented, real road image can be divided into road region and non-road region based line region growing, for lane detection with multiple driveways, lane markings are renewed with method similar to road image segmentation based lane marks edge, marking line can be detected based lane marking shape, finally it applies linear data fitting to find lane markings. It can figure out the way stretching correctly and give warning of vehicle-deviation under the complicated lane conditions.The algorithm proposed by the paper are used to process 200 actual lane image, thus obtaining right rate close to 81.2%, simulated results show that the algorithm is right and effective.更多还原