基于视觉的行车道检测系统研究-计算机应用技术专业论文.docxVIP

STUDY OF THE VISION BASED LANE DETECTION SYSTEM A Dissertation submitted in fulfillment of the requirements of the degree of MASTER OF PHILOSOPHY from Shandong University of Science and Technology by Lin Qing Supervisor: Professor Liang Yongqua n College of Information Science and Engineering June 2010 声 明 本人呈交给山东科技大学的这篇硕士学位论文，除了所列参考文献和世所 公认的文献外，全部是本人在导师指导下的研究成果。该论文资料尚没有呈交 于其它任何学术机关作鉴定。 硕士生签名： 日 期： AFFIRMATION I declare that this dissertation, submitted in fulfillment of the requirements for the award of Master of Philosophy in Shandong University of Science and Technology, is wholly my own work unless referenced of acknowledge. The document has not been submitted for qualification at any other academic institute. Signature: Date: 山东科技大学硕士学位论文摘要 山东科技大学硕士学位论文 摘要 摘 要 基于图像采集和处理技术，研究了一种可用于智能车辆中的行车道检测系统。系统 由图像采集和图像信息处理两个部分组成。图像的采集由架设在智能车挡风玻璃正上方 的 CCD 摄像机完成。采集得到的道路视频图像随后输入图像信息处理系统进行处理， 处理的目标是从从图像中检测车道线的位置，并识别车道线的类型。图像信息处理算法 由单帧图像处理和多帧帧间图像信息处理两个级别构成。在单帧图像处理中综合使用车 道线的边缘、方向、宽度、颜色等特征来对车道线进行识别，并设计了基于网格的形态 学算法、扩展边缘链算法、边缘链距离扫描算法来提取车道线特征点。在提取到车道线 特征点以后，提出了基于直线段的车道线模型和分段霍夫变换来进行车道线的拟合，并 根据拟合直线长度与车道线边缘链长度的比率设计了贝叶斯分类模型来对车道线的虚实 线型进行分类。在多帧帧间图像信息处理部分，利用对单帧中检测到的车道线参数历史 数据的累积统计和分析，实现了车道线参数的帧间优化与跟踪。 关键词：车道线检测，车道线跟踪, 基于网格的形态学，边缘点链接，分段霍夫变 换 Abstract This paper presents a real time vision-based lane detection method to detect lanes by using an on-board monocular camera fixed at the front of the vehicle. In the proposed method, two levels of processing are involved. One is a single frame level, one is an inter-frame level. At the single frame level, after down- sampling the video image, a grid-based morphology operation is firstly used to generate regions of interest(ROI), in which lane-marks are searched for by scanning for an edge-link pair group that corresponds to one lane-mark. During this process, an extended edge-linking algorithm with directional edge-gap closing is used to produce more complete edge-links, and features like lane-mark edge orientation and lane-mark width are used to s