Gradient Correlation Functions in Digital Image Correlation

Abstract

The performance of seven different correlation functions applied in Digital Image Correlation has been investigated using simulated and experimentally acquired laser speckle patterns. The correlation functions were constructed as combinations of the pure intensity correlation function, the gradient correlation function and the Hessian correlation function, respectively. It was found that the correlation function that was constructed as the product of all three pure correlation functions performed best for the small speckle sizes and large correlation values, respectively. The difference between the different functions disappeared as the speckle size increased and the correlation value dropped. On average, the random error of the combined correlation function was half that of the traditional intensity correlation function within the optimum region.

Comparison of Digital Image Correlation Methods and the Impact of Noise in Geoscience Applications

“…where IN x , IN y are the first derivatives in the two directions. A GR image should be insensitive to changes of image intensity and to some extent, to changes of illuminations [20], but it could be more affected by noise [25]. The image orientation [26] is defined as:…”

“…Sjödahl [25] used synthetic and real laser speckle images with different speckle size to compare the performances of seven correlation functions. He considered NCC applied to image intensity (CI), gradient (G) and Hessian (H).…”

Special Issue on Advances in Digital Image Correlation (DIC)

“…The choice of correlation metrics itself is still under investigation. For instance, in [15], the authors present different metrics based on the gradients of the image rather than on the grey level. The above-mentioned question of large deformations implies, in addition to experimental constraints, particular complexities from an algorithmic point of view.…”

Digital Image Correlation Method for Analysis of Reinforced Concrete Structures

“…Although there is the number of obvious advantages of Digital Image Correlation Method, there are still certain issues, which need to be considered when choosing such approach for experimental investigation. Thus, a number of scientists of recent years have conducted studies tended to development of DIC method with the purpose of increasing its accuracy [28][29][30]. For example, the study of Forsey A. and Gungor S., 2016 [28] was aimed to identify the most appropriate technique for images converting.…”

“…It was shown that the bicubic spline based in-house demosaicing method outperformed the other methods in terms of accuracy and aliasing suppression. Sjödahl M., 2019 in his work [29] makes an attempt to improve the method by investigation the new complex correlation function as combination of the pure intensity correlation function, the gradient correlation function and the Hessian correlation function and managed to reduce the random error by half. According to Tomicevic Z. et.…”