超声相控阵无损检测图像多目标跟踪算法  

Multi-target Tracking Algorithm for Ultrasonic Phased Array NDT Images

作　　者：王文强 马立武 刘永洛 WANG Wen-qiang;MA Li-wu;LIU Yong-luo(College of Mechanical and Automobile Engineering;College of AirTransportation, Shanghai University of Engineering Science)

机构地区：[1]上海工程技术大学机械与汽车工程学院 [2]上海工程技术大学航空运输学院

出　　处：《化工自动化及仪表》2019年第1期67-71,共5页Control and Instruments in Chemical Industry

摘　　要：提出一种超声相控阵无损检测图像的多目标跟踪算法。该算法能自动检测和跟踪多个弱小目标,无需任何学习过程和前景目标大小、形状等先验知识。通过Phasor-XS超声相控阵检测仪对试验模板采集的图像序列进行多次跟踪试验验证,结果表明:该方法能有效地对缺陷区域进行跟踪。In this paper, a multi-target auto-tracking algorithm for ultrasonic phased array NDT images was proposed. The method can automatically detect and track several weak targets and doesn’t ask for any learning process and the prior knowledge of the size and shape of the foreground object. Employing the Phasor-XS ultrasonic phased array detector to verify the multiple tracking test of the image sequence of the experimental template detection shows that, this method can track the defect area effectively.

关 键 词：无损检测 多目标跟踪 均值漂移 背景差分法 

分 类 号：TH865[机械工程—仪器科学与技术]