Research Of Digital Image Correlation Method Based On Image Feature Matching Technology

Digital image correlation(DIC) is a product of the combination of modern computer technology and image processing technology. It has been used as a non-contact measurement tool to measure the deformation of objects. Its basic principle is use the correlation criterion function to calculate the deformation of the object’s surface speckle image and obtain the displacement and strain. With the advantages of full-field and non-contact measurements, simple hardware, low environmental requirements, high efficiency, resolution and range easy to adjustment and developed about thirty years, the DIC have been used in modern engineering measurement widely. In order to meet the modern measurement technology requirements of real-time and high accuracy, explore the ways to improve the calculation accuracy and efficiency of the digital image correlation method has been the research focus.The main factors of influence the calculation accuracy and efficiency of the DIC are includes: hardware environment of acquire sample images, correlation function, shape function and optimization function (or relevant search algorithm). This article mainly search from the aspects of the DIC theory, that is the correlation function, shape function and the relevant search algorithm, to get a new method to improve the calculation accuracy and efficiency of DIC. While the research on the relevant search algorithm is the key point in the DIC research fields, this paper mainly discuss the Newton-Raphson (N-R) which can get a higher accuracy in the sub-pixel displacement measurement. However, the N-R iterative optimization requires accurate initial guess of the unknown parameter. We hereby present an automated and reliable initialization method which utilizes image feature matching. That is:Feature points are first detected in the images with high repeatability and each feature is characterized by a descriptor which is insensitive to common image transformations. The features are then matched across images based on the descriptor similarity and a geometric constraint. The deformation parameter of a point of interest is initially estimated from the affine transformation fitted to the matched features inside the subset area. Finally, the estimated deformation parameter are used as the initial guess and sufficiency accurate to enable correct convergence of the N-R method and obtain the accurate displacements with a extremum result of correlation function.In the experiment, the traditional DIC and the new DIC which based on SIFT (scale-invariant feature transform) and SURF (Speeded Up Robust Features) are used in the deformation measurement of carp scales stretching. The calculation results shown that the traditional DIC method can not calculate the large displacement of points of interest with quickly and robustly. While the DIC method based on SURF and SIFT feature matching presented in this article can make the compute more accurately and rapidly.