Influence of the implementation order of distortion correction and subset matching on digital image correlation

Abstract

The shape function and lens distortion affect the accuracy of digital image correlation (DIC). Many studies have been conducted on the shape function and lens distortion correction, and an appropriate shape function with lens distortion correction can reduce measurement errors. However, the influence of the implementation order of correcting lens distortion with known distortion parameters and matching subsets with different shape functions has not been discussed. The two implementation orders are: correcting matched grid points after matching distorted subsets and correcting distorted images before matching the subsets. The former will cause undermatching, and the latter will produce additional interpolation errors. Simulations and an experiment with known lens distortion parameters were performed to evaluate the accuracy of the two implementation orders. The forward additive Gauss–Newton algorithm with first-order and second-order shape functions was applied to the measurements. The results show that the order of correcting distorted images before matching subsets has higher accuracy as compared with another order, and a second-order shape function should always be used unless the underlying deformation is known to be lower order.