基于双目视觉的三维振动测量方法研究  

Research on 3D vibration measurement method based on binocular vision

作　　者：孙猛超 李铁军[1] 史家浩 李赛雷 郑凯文 刘今越[1] SUN Mengchao;LI Tiejun;SHI Jiahao;LI Sailei;ZHENG Kaiwen;LIU Jinyue(School of Mechanical Engineering,Hebei University of Technology,Tianjin 300131,China)

机构地区：[1]河北工业大学机械工程学院,天津300131

出　　处：《传感器与微系统》2022年第10期26-29,共4页Transducer and Microsystem Technologies

基　　金：国家重点研发计划资助项目(2019YFB1302702);国家自然科学基金资助项目(U1813222);河北省教育厅重点资助项目(ZD2018220)。

摘　　要：为了解决智能巡检机器人对机电设备的状态检测问题,针对双目视觉测量三维振动时存在需要物理目标附着、实时性差和特征点假匹配等问题,提出一种基于特征点的三维振动信息提取的方法。利用改进的特征点检测和匹配算法来追踪目标像素位移。该算法不需要在测量目标上粘附物理标识,通过追踪被测目标像素特征点对被测目标进行实时的三维振动测量。最后搭建了三维振动测量系统。实验结果表明:与传统的接触式三轴加速度计的幅频值对比获得99.97%以上的准确率,与激光测振仪在位移为40μm左右的波形下重合度达到95.15%以上。In order to solve the problem in state detection of electromechanical equipment by intelligent inspection robot, that when measuring three-dimensional(3 D)vibration with binocular vision, there are problems such as the need for physical target attachment, poor real-time performance, and false matching of feature points, a method of 3 D vibration information extraction based on feature points is proposed.An improved feature point detection and matching algorithm is used to track the target pixel displacement.The algorithm does not need to attach physical marks to the measured target, and can perform real-time three-dimensional vibration measurement of the measured target by tracking the pixel feature points of the measured target.Finally, a three-dimensional vibration measurement system was built.Experimental results show that compared with the amplitude and frequency value of the traditional contact three-axis accelerometer, the accuracy rate of the algorithm is more than 99.97 %,the degree of coincidence with laser vibrometer is over 95.15 % under the waveform with a displacement of about 40 μm.

关 键 词：三维测振 特征点 双目视觉 

分 类 号：TP242.[自动化与计算机技术—检测技术与自动化装置]