3D digital image correlation implementation with OpenCV

• Main Author: Qi, Yiru
• Other Authors: Qian Kemao
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/66997
• Institution: Nanyang Technological University

The Digital Image Correlation (DIC) technique has been increasingly popular through the last decade since its first introduction in 1970s. It has seen high level of accuracy and feasibility of DIC techniques in measuring material deformation, displacement and optical motion, so it is especially widely applied in the field of experimental mechanics of materials. The techniques of 2D DIC measure the material deformation by using correlation functions to match the reference and target sub-image, and then applying shape functions to determine transformation parameters. By selecting suitable correlation and shape functions to fit in specific requirements in different experiment situations, the empirical results can be impressing. However, the 2D DIC techniques are only capable of measuring displacement within the plane perpendicular to the camera’s optical axis, and the results are subject to noises generated by the object surface. In addition, if the 3D movement of the object is to be detected, the 2D DIC techniques cannot be applied in this case. Instead, the 3D DIC techniques which take in the application of stereo vision concept should be exploited. In this report, the implementation of 3D DIC techniques using OpenCV library is introduced in detail according to the three stages of the process: camera calibration, stereo matching, and 3D reconstruction. The implementation process is explained by OpenCV function usage, and the output of the program is displayed, followed by the possible limitations and future work of the project. It is revealed in the report that 3D DIC implementation is feasible by using C++ with OpenCV support, and the results are satisfying visually to current implementation stage, but these measurements are only applicable for object surface. Further implementation would need hardware which is capable of detecting internal structure of the materials in order to measure the deformation in terms of voxels.