基于机器视觉的障碍导航辅助系统
发布时间：2019-05-17 07:54
【摘要】：障碍导航辅助系统对智能交通系统的发展以及对色盲色弱者具有十分重要的作用,同时也是无人车智能车研究的一个重点难点之一。如何正确无误的进行辅助导航,这对未来智能交通系统的发展具有十分重要的意义。目前,针对障碍导航辅助系统存在的误检率高,导航失效以及满足不了鲁棒性要求等缺陷,本文提出一种基于双目视觉简单的障碍导航辅助系统,该系统包括两个部分:交通灯的定位与识别,道路可行区域的检测。本文首先通过CCD摄像机采集图像,通过非线性变换将采集到的图像从RGB色彩空间的转换成HSV色彩空间。对于交通信号灯的识别,依据HSV空间中H与V相互独立性以及不同色彩的H值与V值对图像进行分割提取符合条件的候选区域,同时对采集到的源RGB图像进行灰度化并进行灰度形态学的操作等一系列预处理及Hough变换操作对可能存在红绿灯进行定位。经Hough变换操作之后的图像与提取的候选区域进行图像融合,可提取出红绿灯位置并对其进行识别。针对道路可行区域的检测,通过实验验证比较传统算法和智能算法在道路可行区域检测方面存在的一些问题,并对出现的问题和根据道路的基本特征进行分析,提出基于亮度二维图像与大津阈值结合的方法,检测道路可行区域。实验结果表明该方法不仅能够克服阴影,小水域等虚拟障碍物的影响,而且能够正确划分出道路可行区域。实验通过对采集的多幅图像进行了实验仿真,结果表明,该系统对交通灯的正确检测率达到93%,而且能够对道路的可行区域进行正确的划分,并满足时效性要求。
[Abstract]:Obstacle navigation auxiliary system plays a very important role in the development of intelligent transportation system and the weak color, and it is also one of the key difficulties in the research of unmanned vehicle intelligent vehicle. How to correctly carry out auxiliary navigation is of great significance to the development of intelligent transportation system in the future. At present, aiming at the defects of obstacle navigation auxiliary system, such as high error detection rate, navigation failure and unable to meet the requirements of robustness, this paper proposes a barrier navigation auxiliary system based on binocular vision. The system consists of two parts: the location and identification of traffic lights and the detection of feasible areas of roads. In this paper, the image is collected by CCD camera, and the collected image is converted from RGB color space to HSV color space by nonlinear transformation. For the recognition of traffic lights, the candidate regions that meet the conditions are segmented and extracted according to the independence of H and V in HSV space and the H and V values of different colors. At the same time, a series of preprocessing and Hough transformation operations are carried out to locate the traffic lights, such as grayscale and gray morphology of the collected RGB images. After Hough transform, the image is fusion with the extracted candidate region, and the traffic light position can be extracted and identified. Aiming at the detection of the feasible area of the road, some problems existing in the detection of the feasible area of the road are verified and compared by experiments, and the problems and the basic characteristics of the road are analyzed. A method based on the combination of brightness two-dimensional image and Dajin threshold is proposed to detect the feasible area of the road. The experimental results show that the method can not only overcome the influence of shadow, small water area and other virtual obstacles, but also correctly divide the feasible area of the road. The experimental results show that the system has a correct detection rate of 93% for traffic lights, and can correctly divide the feasible areas of the road, and meet the timeliness requirements.
【学位授予单位】：河南科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;TP391.41
本文编号：2478927
[Abstract]:Obstacle navigation auxiliary system plays a very important role in the development of intelligent transportation system and the weak color, and it is also one of the key difficulties in the research of unmanned vehicle intelligent vehicle. How to correctly carry out auxiliary navigation is of great significance to the development of intelligent transportation system in the future. At present, aiming at the defects of obstacle navigation auxiliary system, such as high error detection rate, navigation failure and unable to meet the requirements of robustness, this paper proposes a barrier navigation auxiliary system based on binocular vision. The system consists of two parts: the location and identification of traffic lights and the detection of feasible areas of roads. In this paper, the image is collected by CCD camera, and the collected image is converted from RGB color space to HSV color space by nonlinear transformation. For the recognition of traffic lights, the candidate regions that meet the conditions are segmented and extracted according to the independence of H and V in HSV space and the H and V values of different colors. At the same time, a series of preprocessing and Hough transformation operations are carried out to locate the traffic lights, such as grayscale and gray morphology of the collected RGB images. After Hough transform, the image is fusion with the extracted candidate region, and the traffic light position can be extracted and identified. Aiming at the detection of the feasible area of the road, some problems existing in the detection of the feasible area of the road are verified and compared by experiments, and the problems and the basic characteristics of the road are analyzed. A method based on the combination of brightness two-dimensional image and Dajin threshold is proposed to detect the feasible area of the road. The experimental results show that the method can not only overcome the influence of shadow, small water area and other virtual obstacles, but also correctly divide the feasible area of the road. The experimental results show that the system has a correct detection rate of 93% for traffic lights, and can correctly divide the feasible areas of the road, and meet the timeliness requirements.
【学位授予单位】：河南科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;TP391.41
【参考文献】
相关期刊论文 前6条
1 杨唐文;王敏杰;秦勇;;融合图像边缘和区域特征的道路检测算法[J];东南大学学报(自然科学版);2013年S1期
2 梁云;徐东风;王栋;;结合灰度均衡和改进遗传算法的图像阈值分割[J];计算机工程与应用;2011年16期
3 何海涛;宋健;;基于链码趋势表的交通灯形状识别方法[J];计算机工程;2011年15期
4 徐丽珍;孙霖;;基于AdaBoost的人行横道红绿灯自动识别算法[J];计算机应用与软件;2010年03期
5 柏猛;李敏花;;基于图模型的道路检测方法[J];模式识别与人工智能;2014年07期
6 童刚;朱鸿博;;图像处理在红绿灯判别中的应用[J];微型机与应用;2013年04期
，本文编号：2478927
本文链接：https://www.wllwen.com/kejilunwen/daoluqiaoliang/2478927.html
