基于计算机视觉的拉索振动测试

A Study on Computer Vision-based Cable Vibration Measurement

提出了一种基于计算机视觉目标跟踪算法的非接触式拉索振动测试方法。首先通过目标跟踪算法对相机采集的拉索振动视频图像序列进行分析,识别拉索表面标靶在图像平面的像素位移时程数据,然后根据快速傅里叶变换算法获取振动频谱,采用直接峰值法提取频谱峰值,得到拉索各阶自振频率。介绍了视觉测振方法流程,针对当前视觉测振方法采用的目标跟踪算法鲁棒性差、实时性低的问题,引入了高效的核相关滤波跟踪算法,基于KCF和CN两种经典核相关滤波跟踪算法的原理和MATLAB平台设计了视觉测振系统软件。在实验室中开展了钢绞线振动试验,将视觉测振方法与加速度计方法的频率识别结果进行对比,结果表明:视觉测振方法识别的钢绞线各阶自振频率与加速度计结果非常接近,最大频率相对误差为-0.12%,验证了视觉测振方法的可靠性。

A contactless cable vibration measurement technique based on visual object tracking algorithm was proposed in this paper.At first, by analysing image sequences in the video of cable vibration with object tracking algorithm, It could get dynamic pixel displacement of the artificial target on cable surface.Then, the fast fourier transform was used to calculate the spectrum of cable vibration, and the natural frequencies of cable was identified by the peak-picking method in frequency-domain.The process of vision-based method for vibration measurement was described in detail, and two kernelized correlation filtering alogrithm, KCF and CN,were introduced to improve tracking accuracy and efficiency.A simple MATLAB software to measure vibration of cable based on the principle of the two algorithm was also designed.In order to verify the reliability of the vision-based method, a laboratory vibration test of steel strand was carried out.The results show that the maximum error in the natural frequencies of steel strand, which were obtained using the data of computer vision-based method, relative to those obtained from the accelerometer, was-0.12%.Vision-based technique is a promising research direction and could be further developed to build an automatic and intelligent system for stay-cable vibration measurement.