基于图像的车道线检测与跟踪

\

2013届毕业生毕业设计说明书

题目: 基于图像的车道线检测与跟踪系统

院系名称：信息学院专业班级：计科0905班学生姓名：王曌盟学号： ************ 指导教师：于俊伟教师职称：讲师

2013年5月28日

基于图像的车道线检测与跟踪系统

摘要

随着我国经济的不断发展和科学技术的不断进步以及生活水平的不断提高，汽车正逐步成为大众的交通工具。随着汽车的不断普及以及汽车行车速度的不断提高，交通事故的数量也随之上升。每年的交通事故给国家的经济，人民的生命和财产造成了巨大的损失。由于疲劳驾驶和注意力不集中，引起车辆偏离车道线从而造成交通事故的数量约占所有交通事故的三分之一左右。

在车道偏离预警系统中，整个系统的关键是车道线的正确提取与识别。车道线的提取主要完成从图像中检测并识别出车道线，并且确定车辆在道路上的安全可行区域和定位车道线相对于车辆的位置，以便监测车辆行进的实时情况。当车辆发生偏离时，能够提醒驾驶员及时调整车辆的状态，从而避免交通事故的发生。

本文提出了一种实现对图像中的车道线进行识别，得出完整的车道线信息和检测两条车道线夹角的图像处理方法。本文主要研究的内容有：

第一、对采集的道路图像进行预处理，主要包括图像的滤波、图像的阈值分割，联通域标记，边缘提取等操作。

第二、给出了基于感兴趣区域和Hough变换的车道检测算法，完成对车道线的检测。

第三、对检测出来的车道线建立高斯混合模型，从而预测以后车道线检测的可靠性，并将检测结果与高斯混合模型的预测结果结合得到最接近真实情况的结果。

关键词：车道线检测， Hough变换，高斯混合模型

The detection and tracking of the lane line

based on the image

Abstract

With the continuous development of China's economy and the continuous progress of science and technology as well as the continuous improvement of living standards, the car is gradually becoming a public transport. With the growing popularity of the automobile, as well as the continuous improvement of automobile traffic speed, the number of traffic accidents also increased. Annual traffic accidents caused huge losses to the national economy and people's lives and property.Causing the vehicle deviates from the lane line resulting in the number of traffic accidents accounted for about one-third of all traffic accidents due to driver fatigue and inattention.

In the lane departure warning system, the key to the whole system is the extraction and recognition of the right of the lane line. Lane line extraction completed to detect and identify the lane line from the image, and determines the feasible region of the vehicle on a road safety and positioning of the lane line position relative to the vehicle, in order to monitor the vehicles traveling in real-time situation. When the vehicle deviates able to alert the driver to adjust the state of the vehicles in a timely manner, so as to avoid traffic accidents.

The paper presents an implementation of the lane mark in the image recognition, to draw the complete information of the lane mark and the image processing method of detecting the angle between two lane line. This paper studies the content:

First, on the acquisition of road image preprocessing, including image filtering, image thresholding segmentation the Unicom domain mark, edge extraction operation.

Second, given the lane detection algorithm, to complete the detection of the lane line based on the region of interest and Hough transform.

Detected lane lines to establish Gaussian mixture model in order to predict the