Efficient digital image correlation using gradient orientation

Conventional digital image correlation (DIC) searches the target point in the searching area by computing the pointwise correlation coefficients and this process is time consuming. In this paper, we propose to use the gradient orientation (GO) technique to find the potential target point quickly and compute the correlation coefficients for the potential target points only. So the coarse searching efficiency is enhanced. Compared with gray intensity, it is proven that GO is invariant to illumination variation. If the cosine value of the gradient orientation error (GOE) between the GO at the seed point and the GO at the searching point is relatively large, the searching point is the potential target point and we need to compute its correlation coefficient to make further judgement; otherwise, it can be rejected directly. Additionally, in order to improve the robustness of the proposed DIC, the original seed points are corrected based on the following criterion: (1) the gradient of the seed point and of at least two eight-neighborhood points should be relatively large; (2) the GO of those at least two eight-neighborhood points should be close to the GO of the seed point. Experiments are carried out to validate the proposed DIC. It is shown that the rate of the number of potential target point is less than 0.2 and the coarse searching efficiency is improved by more than 50% without reducing the measurement accuracy. (C) 2018 Elsevier Ltd. All rights reserved.