摘要道路边界检测是自主车系统的一个重要部分，乡村道路的边界检测较一般的道路边界检测相比更具挑战性。通常，道路检测一般采用单一的可见光传感器，导致乡村道路边界检测常常不能应对全天时的道路环境，本文利用可见光传感器和红外传感器同时进行乡村道路的检测，依据两种图像的特点，分别采用了道路图像的预处理、道路的消失点检测，边界假设的生成和真实道路边界的选择方法分，最后，将两种图像所得到的道路边界分别利用标定参数变换到三位坐标系进行拟合。利用真实无人车拍摄的图像进行实验，证明本文提出的方法是有效的。64830

毕业论文关键词  边界检测  可见光图像  红外图像   消失点检测

毕业设计说明书（论文）外文摘要

Title      Research on the Algorithm of Edge Detection  Country Road Based on Bi-Visual Sensors                                                

Abstract The edge detection of roads is a significant part of autonomous land vehicles. The edge detection of country roads is more challenging than the edge detection of common roads. Usually, only one type of sensor is applied in the process, which is visible-light-sensor. Therefore, the system is not available during the whole day. This essay applies two types of sensors, which are visible-light-sensor and infrared-sensor, for the edge detection of country roads. According to the characteristics of the two types of images, image-preprocessing and vanishing point detection are applied respectively. In addition, edge assumption and real edge selection are used to find the real edges. Finally, the results from both sensors are combined in the three-dimensional system of coordinate, using straight line fitting. The material that is used in this essay is collected by real autonomous land vehicles, so the method which is proposed in the essay is proved to be valid.

Keywords  Edge Detection  Visible-light Image  Infrared Image        End Point Detection

目   录

1  绪论 1

1．1  课题简介 1

1．2  研究意义 1

1．3  国内外研究现状 2

2  可见光道路检测 4

2．1  消失点检测 4

2.1.1  Gabor变换 4

2.1.2  投票机制 6

2.1.3 消失点的确认 7

2．2  图像预处理 9

2.2.1  图像平滑（去除噪声） 9

2.2.2  边缘检测 10

2．3  边界假设与选择 12

2.3.1  将彩色图像转换为灰度图像 12

2.3.2  边界假设 12

2.3.3  对应一条假设边界的数据采集 15

2.3.4  假设边界的评价与选择 17

2.3.5  空出缓冲带的处理点的采样 19

3  红外图像边界检测 22

3．1  消失点检测 22

3．2  图像预处理