Experimental analysis of image noise and interpolation bias in digital image correlation

The popularization of the digital image correlation (DIC) method has raised urgent needs to evaluate the accuracy of this method. However, there are still some problems to be solved. Among the problems, the effects of various factors, such as the image noise caused by the camera sensors, the employed interpolation algorithm, and the structure of the speckle patterns, have become a major concern. To experimentally measure the position-dependent systematic error (i.e. interpolation bias) caused by non ideal interpolation algorithm is an important way to evaluate the quality of the speckle patterns in the correlation method, and remains unsolved. In this work, a novel, simple and convenient method is proposed to measure the interpolation bias. In the new method which can avoid the out-of-plane displacements and the mechanical errors of translation stages, integral-pixel shifts are applied to the image shown on the screen so that sub-pixel displacements can be realized in the images captured by the camera via proper experimental settings. Besides, the fluctuations of the image noise and the sub-pixel displacement errors caused by the image noise are experimentally analyzed by employing three types of cameras commonly used in the DIC measurements. Experimental results indicate that the fluctuations of the image noise are not only proportional to the image gray value, but also dependent on the type of the employed camera. On the basis of eliminating the image noise via the image averaging technique, high precision interpolation bias curves more than one period are experimentally obtained by the proposed method. (C) 2016 Elsevier Ltd. All rights reserved.