基于机器视觉的管道通径内检测技术

1. 中国石油管道科技研究中心· 油气管道输送安全国家工程实验室;2. 中国石油管道公司;3. 中国石油大学(北京)机械与储运工程学院

管道;内检测;机器视觉;图像处理;Radon 变换

A pipeline internal diameter detection technology based on machine vision
LI hua1, LIU Yang2, MENG Xiangpeng2, ZHANG Wenxue3, JIN Yuxin2

1. PetroChina Pipeline R & D Center//National Engineering Laboratory for Pipeline Safety; 2. PetroChina Pipeline Company; 3. College of Mechanical and Transportation Engineering, China University of Petroleum (Beijing)

pipeline, internal detection, machine vision, image processing, Radon transform

备注

管道通径内检测可以提前发现管道安全隐患,在确保管道安全运行方面发挥着重要作用。在机器视觉技术的基础上,将视觉测量技术中的Radon 变换应用于激光光源投射成像方法中,提出了基于视觉的管道通径测量计算方法。该方法利用激光发生仪投射与管道内部形状完全吻合的激光环,由相机采集管壁上的光环图像,并将其进行边缘检测和细化处理,最终获得管道的截面轮廓曲线。通过Radon 变换得到两条反映管道内径信息的光带,经细化处理提取光带中心线,进而根据中心线之间的距离计算得到管道在所有方向角上的内径数据。研究结果表明:该测量计算方法高效快捷,能够一次性获取管道截面在所有方向角上的内径信息,具有很高的实际应用价值。(图7,参[21]

The pipeline internal diameter (ID) detection technology plays an important role in discovering pipeline hazards in advance and ensuring safe operation of pipelines. In this paper, a vision-based pipeline ID measurement and calculation method was developed by applying the Radon transform of vision measurement technology to the laser-source projection imaging method based on the machine vision technology. In this method, the laser transmitter is used to produce a ring which is completely in line with the internal shape of the pipeline. The image of the laser ring on the wall of the pipeline is collected by a camera and then transmitted to the computer for edge detection and refining treatment. And eventually, the cross section profile of the pipeline is obtained. What’s more, two optical bands which reflect the information of pipeline ID are acquired by virtue of the Radon transform, and their center lines are extracted by means of refining treatment. And then, the pipeline ID data at all directional angles are calculated according to the distance between the center lines. It is indicated that this vision-based pipeline diameter measurement and calculation method is efficient and fast with high value of practical application, and it can obtain the information of pipeline ID at all directional angles at one time. (7 Figures, 21 References)