基于主动式全景视觉的管道形貌缺陷检测系统  

Pipe morphology defects inspection system with active stereo omnidirectional vision sensor

作　　者：汤一平[1] 鲁少辉[1] 吴挺[1] 韩国栋[1] 

机构地区：[1]浙江工业大学信息工程学院,浙江杭州310023

出　　处：《红外与激光工程》2016年第11期176-182,共7页Infrared and Laser Engineering

基　　金：国家自然科学基金(61070134)

摘　　要：针对现有的管道缺陷检测技术不能同时对管道的形、貌缺陷进行检测与评估这一工程难题,在前期研究工作的基础上,设计了一种基于主动式全景视觉的管道内部缺陷检测系统,能够快速获取管壁密集点云的三维坐标,同时对内壁表面缺陷进行检测与评估。首先利用主动式全景视觉传感器(AODVS)实时获取内壁全景图像和激光横断面扫描全景图像,然后对管道内壁全景图像进行柱状展开、预处理和缺陷检测及分类等处理;然后对激光横断面扫描全景图像处理,计算管道内壁点云的三维坐标,进一步对管道缺陷部分进行定量分析,最后利用三维建模技术重构带有真实纹理信息的管道模型。实验结果表明:文中设计的检测系统能够对管道凹凸形变、孔洞、管壁裂缝、腐蚀等缺陷进行检测与分析,具有较高的检测精度,为管道内表面三维测量和重构提供了一种新的手段。For the engineering problem of low efficiency of defects inspection and assessment of pipe with existing method, an in-pipe internal inspection system based on active stereo omnidirectional vision sensor（AODVS） was presented to acquire 3D coordinates of point cloud and detect the defects on the inner surface of pipes in real time. First, inner surface panoramic images and laser streak panoramic images were captured with AODVS, Inner surface images were processed as follow：unwrapping, preprocessing, feature extracting and defects classification, Laser streak images reflecting the shape of inner pipe were processed to calculate 3D coordinates of the point cloud of inner surface. Finally, the pipe＇s triangular grid model with real texture information was reconstructed by 3D modeling technique. Experiment results show the efficiency of proposed method to detect racial variation, holes, cracks and corrosions with high accuracy, this system provide a new online inspection approach to 3D measurement and reconstruction of industrial pipes.

关 键 词：主动式全景视觉传感器 视觉三维测量 全景激光 管道缺陷检测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]