Research On Vibration Feature Extraction Method Based On High-speed Vision

Measurement technology is one of the basic technologies of modern industry.To a certain extent,it reflects the development level of national manufacturing industry and science and technology.Because the traditional contact measurement technology is easily affected by environmental factors,it is difficult to apply in some occasions.With the development of digital image technology and the progress of hardware technology,the accuracy and efficiency of image sensor have been greatly developed,and a new detection technology-visual measurement technology has been produced.In this paper,visual measurement is taken as the research object.Aiming at the problems of low automation and insufficient anti-interference ability of traditional image processing algorithms,three image processing methods are introduced,and three displacement extraction algorithms are proposed based on these three image processing methods.Firstly,a FAST feature point displacement extraction algorithm is proposed,which mainly improves the tracking efficiency of FAST feature points by reducing the number of pixels compared with candidate points.This algorithm is applied to the vibration measurement experiment of hacksaw cantilever beam,and it is verified that FAST displacement extraction algorithm has higher displacement extraction speed.Then,aiming at the problem of rigid body motion of objects,an improved zero-mean normalization sum of squared differences(ZNSSD)template matching algorithm is proposed,that is,the image pyramid decomposition algorithm and surface fitting algorithm are combined with ZNSSD template matching algorithm to improve the accuracy and speed of the traditional ZNSSD template matching algorithm.Through the experiment of grating ruler mobile platform,it is verified that the improved ZNSSD template matching algorithm has higher accuracy and speed.Finally,aiming at the problem of extracting non-rigid motion vibration signals of objects,a data dimensionality reduction method based on singular value decomposition(SVD)is introduced.The algorithm is applied to extract the trajectory of the simulation circle,and it is verified that the SVD decomposition algorithm has the function of extracting displacement information.At the same time,aiming at the problem that the displacement signal extracted by SVD decomposition algorithm contains a lot of noise,the iterative SVD denoising method and the improved wavelet denoising method are further proposed,and the effectiveness of the two methods is verified by simulation.Finally,the proposed visual measurement method is integrated into the built high-speed visual measurement system,and the system is used to measure and analyze human respiratory signal and grating ruler movement.The experiment proves that the visual measurement technology is an effective measurement method,and provides a reliable method for the development of measurement technology.