Algorithm of crack tracking during fatigue test through calculating the optical flow

Abstract

An algorithm of crack detection during fatigue testing of materials, designed to automate the process of cyclic loading and tracking the crack tip, was proposed and tested. The ultimate goal of the study is aimed at controlling the displacements of the optical system with regard to specimen under fatigue loading to ensure observation of the area of interest. It is shown that the image region that contains the crack may be detected and positioned with an average error of 1.17% for noisy images and 11.5% when blurred images are analyzed. In terms of determining the crack tip position. the algorithm provides the accuracy of its localization with the average error value of 9.23 pixels when processing noisy images, while it makes 42.7 pixels for blurry ones.