Fast initial value estimation in digital image correlation for large rotation measurement

Abstract

Digital image correlation (DIC) is widely used in macroscopic and mesoscopic mechanical tests because of its advantages of non-contact, high precision, full-field measurement and simple experimental equipment. The application of various nonlinear optimisation algorithms greatly reduces the computation time of the DIC iteration process. Thus, efficiently obtaining a reliable initial value is crucial. Particularly, in cases when the surface of the test object is substantially rotated or the deformation involves a large rotation, the initial value estimation can have a major influence on the execution speed of the algorithm. Some scholars have proposed initial value estimation methods for large-rotation objects, but they all sacrificed the speed of calculation. This study deals with improving the efficiency of the initial value estimation algorithm for large-rotation objects from two aspects. Firstly, we decomposed deformed and reference images into multiresolution layers to create wavelet pyramids, and the correlation coefficients between the two compared images were calculated at the low-resolution layer. The multiresolution processing of the image data provides an efficient method for registering large image data sets because the full-size data sets does not require matching. Secondly, we propose a local ring pattern (LRP), which is invariant to object rotation, to convert the 2D template into a 1D grey value sequence for calculating the correlation coefficients. The advantages of the LRP feature include the characterisation of its rotation invariance and the reduction of computational complexity. Experimental results indicate that the proposed method can be used to estimate an initial value where reference and deformed subsets are related by translational and rotational motions and the speed of the proposed method is higher than that of the traditional method.