基于图像处理的交通锥识别与定位算法研究

发布时间：2018-05-24 02:45

本文选题：交通锥 + 图像处理　；参考：《郑州大学》2017年硕士论文

【摘要】：本文从数字图像处理角度出发,对交通锥的识别与定位算法进行了深入研究。该算法主要用于辅助交通锥收放车对交通锥进行自动识别与定位,有利于提高交通锥收放车的收锥效率和自动化性能。论文主要内容包括以下三个方面:(1)依据交通锥的颜色和形状特性,提出了一种基于连通域大小和特征点位置关系的识别算法。首先基于HSV颜色空间对获取的交通锥图像进行分割,然后利用形态学方法对颜色分割后的二值图像进行一系列处理,包括基于连通域大小的目标筛选、轮廓提取和基于轮廓外接矩形特征点的分析,以实现交通锥的识别。(2)基于几何关系推导法,对交通锥的定位进行了建模。依据交通锥红色区域的特征点,根据摄像机的投影模型,结合它们之间的几何关系,推导出交通锥在路面上的位置信息。(3)根据交通锥的识别与定位结果,为后续机械臂的抓取建立交通锥角度的模板库。基于特征点的位置关系进行交通锥倒下状态时角度的测算,由于交通锥图片是带有高度的摄像头拍摄到的,三维信息在转化成二维图片中交通锥角度的过程中由于缺少深度信息从而造成误差。因此需要建立一个映射的模板库,模板库中包含了从二维图像中的角度到机械臂抓取角度的映射。为了对本文算法进行验证,在Visual Studio 2010+OpenCV平台上开发了相应的应用程序,对以上的研究内容进行了较为系统的实验测试。实验结果表明,该算法能够有效辨认出路面上的交通锥,实验中交通锥的识别率为97.8%,交通锥的定位误差均小于等于2厘米,交通锥角度的测量误差小于16°,能够基本满足交通锥收放车上机械臂对交通锥抓取工作的精度要求。
[Abstract]:From the angle of digital image processing, this paper studies the recognition and location algorithm of traffic cone in depth. This algorithm is mainly used to assist the automatic recognition and positioning of traffic cones for traffic cones. It is beneficial to improve the efficiency and automatic performance of the cones. The main contents of this paper include the following three aspects: (1) According to the color and shape characteristics of the traffic cones, a recognition algorithm based on the relationship between the size of the connected domain and the location of the feature points is proposed. First, a traffic cone image is segmented based on the HSV color space, and then a series of two value images are processed by the morphological method, including the target based on the size of the connected domain. The identification of traffic cones is realized by screening, contour extraction and rectangle feature points based on contour lines. (2) based on geometric relation derivation, the positioning of traffic cones is modeled. According to the feature points of the traffic cone red area, the traffic cone is derived on the road according to the projection model of the camera and the geometric relationship between them. Position information. (3) according to the recognition and positioning results of the traffic cone, the template library of the traffic cone angle is set up for the capture of the following manipulator. Based on the position relation of the feature points, the angle of the traffic cone falls down, because the traffic cone picture is photographed with the height of the camera, and the three-dimensional information is converted into the two dimensional picture. In the process of cone angle, the error is caused by lack of depth information. Therefore, a mapping template library is needed. The template library contains the mapping from the angle of the two-dimensional image to the grasping angle of the manipulator. In order to verify this algorithm, the corresponding application is developed on the Visual Studio 2010+OpenCV platform. The experimental results show that the algorithm can effectively identify the traffic cones on the road, the recognition rate of the traffic cone in the experiment is 97.8%, the positioning error of the traffic cone is less than 2 cm, and the measurement error of the angle of the traffic cone is less than 16 degrees, and it can basically meet the machinery on the traffic conical retractable car. The precision requirements of the arm to catch the traffic cone.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】