Research On Measuring Large Inter-frame Deformation Based On Digital Image Correlation (DIC) Method

The decorrelation phenomenon occurs when the digital image correlation method measures large inter-frame deformation,and the existing algorithms for measuring large inter-frame deformation based on digital image correlation method are still some imperfect place.Based on the existing research results,a digital image correlation full-field initial guess method based on bilateral feature matching is proposed in this paper to measure large inter-frame deformation,which can address the above phenomenon.Firstly,the ORB(Oriented FAST and Rotated BRIEF)fast feature detection algorithm is used to extract and match the feature point pairs of the reference and target images to overcome the decorrelation phenomenon of digital image correlation when measures large inter-frame deformation.The ORB algorithm ensures that evenly distributed matching points are extracted in the full-field even under large deformations,and reliable number of matching feature point pairs are provided for the initial guess of points of interest in the full-field.Then,the grid-based motion statistics method in the field of computer vision is introduced to deal with the problem of false matching in the process of feature matching.To make full use of the advantages of the grid-based motion statistics method and solve the problems of its application in the matching process of large deformation speckle image,the grid motion statistics method is improved.The Gaussian kernel function is used to assign reasonable weights to the matching feature points and the bilateral consistency check limits the rate of false matches,which further improves the robustness,accuracy and computational efficiency of the grid motion statistics method,making it suitable for feature matching and screening of large inter-frame deformation.Finally,a simple,path-independent initial guess method of full-field deformation is used to replace the existing path-dependent full-field measurement method based on digital image correlation.In this process,the MSAC(M-Estimate Sample Consensus)algorithm is used to further eliminate false matches in the search region of each point of interest,and the affine transformation model and the deformation of the point of interest are independently estimated.The inverse compositional Gauss-Newton algorithm is provided with reliable initial guess of the full-field deformation,and the final deformation results are obtained iteratively.Simulations and experiments verify the performance of this method to measure different large inter-frame deformation and its practicability in engineering applications.Experimental results demonstrate that the real deformation can be measured by this method is about 40%.Compared with other algorithms,this method measures the full-field deformation parameters of large inter-frame deformation more efficiently and reliably.