Experimental Study On Measurement Of 3D Surface Vibration Based On Image Analysis

In the field of industrial measurement,obtaining the three-dimensional information reconstruction model of the object surface is the basic step for subsequent analysis and decision,which is also the prerequisite for measuring the vibrating object.Traditional contact measurement method can only measure static objects due to the limitation of measurement method.In order to make fast full-field measurement,non-contact optical measurement method is needed.The fringe projection method and Fourier transform profilometry,with their advantages of fast calculation speed,instantaneous information acquisition and wide application,can meet the requirements of measuring vibrating objects.In this paper,the three-dimensional surface topography of a static object is obtained by Fourier transform profilometry method based on fringe projection,and a low-frequency vibration measurement system based on fringe projection is proposed.Firstly,this paper introduces the necessity of vibration measurement and the method of surface profile measurement,deduces the measurement principles of fringe projection and Fourier transform profilometry,and analyzes the gradient measurement limit of this method in detail.Secondly,by explaining the inevitable phase ambiguity and error accumulation in the phase expansion process of the phase measurement method,Then,combined with the theoretical basis of Fourier fringe profilometry in vibration measurement,a controllable vibration object is measured.By comparing the image acquired by the camera with the reference image,Fourier transform,window filter,inverse Fourier transform and height was measured,and a specific reference point in the height restoration diagram was extracted,and the frequency and amplitude of the vibrating object were analyzed through the height distribution map.After analyzing the sensitivity and resolution of frequency and amplitude measurement using rectangular wave as projection pattern,sinusoidal wave and triangular wave as projection pattern were respectively used to compare the measurement results.Finally,the measurement errors on the measurement process were discussed and analyzed theoretically.The method adopted in the fourth chapter combines the characteristics of Kinect V2 that can collect depth information with the advantages of artificial neural network in processing complex data to realize the automatic detection of vibration frequency of the target object.This method has the advantages of no contact,no marking,using artificial data for network training,no need for pre-processing operations such as denoising and enhancement,automatic detection of vibration frequency,etc.Moreover,due to the characteristics of adding noise to artificial training data,this method has a strong robustness.However,due to the inherent defects in the accuracy of Kinect V2 depth sensor,the noise of electronic and mechanical vibration and camera shake will interfere with the measured depth information.Moreover,because Kinect V2 adopts infrared technology,the change of sunlight irradiation intensity and random exposure will also cause some interference to it.Therefore,this method is currently limited to indoor applications and cannot detect all the frequency components of the measured objects discussed further in this chapter.Nevertheless,our experiment also achieved the expected purpose.Based on the experimental results,we can conclude that the feature extraction ability of neural network can be utilized in the field of optical vibration extraction and processing,which provides new possibilities for the development and research of optical vibration measurement based on deep learning.It can be seen from the comparison between the prediction result of neural network and the peak value of the target vibration depth signal through FFT transformation that the predicted frequency is consistent with the real frequency domain of the target vibration,and it can also be seen from the confusion matrix that the method achieves accurate detection.This paper designs two kinds of non-contact optical image measurement method,the analyte for low frequency vibration measurement study,solves the analyte in the optical image measurement of 3-D surface reconstruction,the further study of the classical Fourier fringe contour measurement method,and explores the new stereo camera depth information convolution neural network,the theory of the technology involved can be used in the environmental measurement and industrial monitoring.