干扰环境下基于计算机视觉的多目标动位移高精度监测方法  

A computer vision-based method for high-precision monitoring of multitarget dynamic displacement in interference environments

作　　者：周洲 陈太聪[1] ZHOU Zhou;CHEN Tai-cong(State Key Laboratory of Subtropical Building Science,School of Civil Engineering and Transportation,South China University of Technology,Guangzhou 510641,China)

机构地区：[1]华南理工大学土木与交通学院,亚热带建筑科学国家重点实验室,广东广州510641

出　　处：《振动工程学报》2021年第5期979-986,共8页Journal of Vibration Engineering

基　　金：国家自然科学基金资助项目(52178479);亚热带建筑科学国家重点实验室自主研究项目(2017KB13);广东省现代土木工程技术重点实验室项目(2021B1212040003)。

摘　　要：当前基于计算机视觉的动位移测量研究通常要求高速高分辨率摄像机和理想拍摄环境,以保证测量的性能和精度。然而高速相机成本较高,目标成像需要较高对比度,且实际拍摄过程中环境条件也难以保持稳定,导致应用受限。结合时空上下文算法和光流算法,提出一种无需人工标靶点、鲁棒的多目标位移监测方法,通过智能手机实现干扰环境下的结构多点动位移同步测量。开展悬臂小球模型的扫频实验,检验方法在一定频率范围内的测量效果。其中,使用智能手机对激振小球进行拍摄,并在实验中保留复杂背景和模拟光照变化。分别采用所提方法和常用的特征光流算法对视频进行处理,得到动位移结果,并与位移传感器测量值进行对比。结果表明,在有光照变化干扰下,所提方法具有更强的抗干扰性,各监测点的最大位移偏差在5%以内。At present,research on dynamic displacement measurement based on computer vision usually requires high-speed and high-resolution camera as well as ideal shooting environment to ensure the performance and accuracy of measurement.However,high cost of the camera,requirements of high image contrast and stable environment during the shooting process limit the wide application of the technology.In this paper,a robust multi-target displacement monitoring method without artificial target is proposed based on the spatio-temporal context algorithm and the optical flow algorithm.The dynamic multi-point displacements of the structure under the interference environment are synchronously measured by smart phone.The frequency sweep experiment of the cantilever sphere model is carried out to test the measurement effect in a certain frequency range.In the experiment,a smart phone is used to shoot the exciting sphere,and the complex background is preserved and the illumination is simulated.Then,the video is processed with the proposed method and the common characteristic optical flow algorithm,and the dynamic displacement results are obtained and compared with the measurement from the displacement meter.The results show that the proposed method has stronger anti-interference ability under the interference of illumination variation.In addition,the maximum displacement deviation of each monitoring point is within 5%.

关 键 词：结构健康监测 位移测量 计算机视觉 时空上下文 光流 

分 类 号：O327[理学—一般力学与力学基础] TU317[理学—力学]