自适应聚乙烯管道超声信号线检测算法  

Adaptive Detection Algorithm for Ultrasonic Signal Lines of Polyethylene Pipeline

作　　者：李礼彬 顾震 周家乐 颜秉勇[1] 王慧锋[1] LI Libin;GU Zhen;ZHOU Jiale;YAN Bingyong;WANG Huifeng(School of Information Science and Engineering,East China University of Science and Technology,Shanghai 200237,China)

机构地区：[1]华东理工大学信息科学与工程学院,上海200237

出　　处：《华东理工大学学报（自然科学版）》2021年第6期726-733,共8页Journal of East China University of Science and Technology

摘　　要：在聚乙烯管道电熔焊接接头的超声成像中,特征线、电阻丝线和管材内壁回波这3条信号线的检测是实现焊接缺陷自动化识别的关键技术。针对超声图像纵向分层显著、全局干扰较多的特点,提出了一种自适应聚乙烯管道超声信号线检测算法。首先,为了识别分层区域,基于形态学运算对灰度图进行自适应模糊,再进行水平均值投影;然后,基于目标特征和空间信息改进了自适应阈值分割算法,用于检测特征线,并提出了Otsu-CRF算法用于检测电阻丝并解决电阻丝粘连问题;最后,基于色彩空间和阈值分割实现管材内壁回波的检测。实验结果显示,本文的自适应算法能有效地检测3条信号线,与已有算法相比提升了检测结果的完整性、准确性和高效性,验证了将该算法用于聚乙烯管道电熔焊接接头的超声信号线自动化检测的可行性。In the ultrasonic imaging of polyethylene(PE)pipe joints,the detection of characteristic line,resistance wire and echo of pipe inner wall is the key technology to realize automatic identification of welding defects.Aiming at the characteristics of obvious longitudinal stratification and more global interference in ultrasonic image,an adaptive ultrasonic signal line detection algorithm for polyethylene pipeline is proposed.Firstly,the grayscale image is adaptively blurred and projected in the horizontal direction to identify the hierarchical area.Secondly,based on the target feature and spatial information,an improved adaptive threshold segmentation algorithm is proposed to detect feature lines.Thirdly,the Otsu-CRF algorithm is proposed to detect the signal line of the resistance wires.Finally,the detection of pipeline inner wall echo is fulfilled by color space conversion and threshold segmentation.It is shown via experimental results that the proposed adaptive algorithm can completely detect three signal lines and enhance the detection effect and efficiency compared with the existing algorithms.It also verifies the feasibility of applying the algorithm to the automatic detection of ultrasonic signal line of polyethylene pipe electrofusion welding joint.

关 键 词：自适应检测 条件随机场 阈值分割 聚乙烯管道 超声成像 

分 类 号：TN911.73[电子电信—通信与信息系统]