Realtime conversion of cracks from pixel to engineering scale using Augmented Reality

Abstract

A key step in image-based crack characterization techniques is pixel to engineering scale conversion. The conversion factor corresponds to Camera-to-Crack-Distance (CCD) and camera obliquity angle. To date, nonstationary image-based crack characterization methods are not designed for real-time unit conversion. This study develops a new crack characterization algorithm for Augmented Reality (AR) headsets; then implements it in Unity-game-software on a computer; and finally deploys and validates it in AR headsets. The proposed algorithm includes: (1) the transformation of pixel information to AR platform; (2) edge extraction utilizing Canny algorithm; (3) crack measurement at the pixel level adopting the horizontal approach; (4) CCD and camera angle measurement using AR orientation capabilities and applying this information for perspective correction; (5) Engineering scale dimensions computation using the results of (3)–(4). A scaler equation for perspective correction streamlines the algorithm toward real-time implementation. The results show maximum errors of 8.45% and 12.05% for laboratory and field experiments, respectively.