With the in-depth research on dynamic deformation measurement technologies, the measurement accuracy requirements for the large displacement deformations and the discontinuous deformation in dynamic deformations are becoming higher and higher. However, the contact-based measurement technologies often fail to meet the requirements due to their shortcomings, such as limited measurement range, inconvenient operation, and low measurement efficiency. Digital image correlation (DIC) for measuring displacement and strain has developed into one of the main techniques for measuring dynamic deformations, owing to its advantages of non-contact, high accuracy, and high efficiency of full-field measurement. To improve the accuracy and efficiency of measuring large displacement deformations and discontinuous deformations using DIC, this thesis focuses on the 3D full-field measurement for dynamic deformations and studies the principle and key technologies of DIC in-depth. Aiming at the high-robustness and high-efficiency application requirements for DIC in dynamic deformation measurements, especially in the measurement of complex dynamic deformations, this research mainly focuses on how to improve the robustness and reliability of the initial displacement calculation, and the accuracy and speed of the subpixel displacement calculation in DIC. The main research content and the results obtained include:

Firstly, a feature-assisted initial displacement calculation method is developed to calculate the accurate and reliable initial displacement value for the large displacement measurement. The DAISY feature descriptor with high computational efficiency and the feature matching algorithm based on PatchMatch are employed to provide full-field initial displacement values for all calculation points, which effectively solves the problem of calculating the initial displacement value under the large displacement. It can provide an accurate and reliable initial displacement value to improve the measurement accuracy and stability of DIC. The simulation experiment results show that this method can provide the initial displacement values for the calculation points while maintaining high accuracy, robustness and reliability.

Secondly, a path-independent subpixel displacement calculation method is proposed and implemented to accurately measure the discontinuous deformation areas in dynamic deformations, which can improve the low measurement efficiency of the path-dependent DIC method. Combined with the feature-assisted initial displacement calculation method, a parallel DIC method is implemented, which independently analyzes the displacement of each calculation point without relying on the transfer path of the initial displacement values for the DIC analysis. This method improves the efficiency and stability of the displacement calculation, which enhances the robustness and efficiency of DIC.

Thirdly, a feature-assisted stereo matching strategy is developed for stereo matching between the left camera and right camera in 3D full-field measurement of complex dynamic deformations. Combined with the epipolar constraint, fast and high-precision stereo matching is realized. Using the high-precision camera calibration method and 3D reconstruction method, a feature-assisted 3D-DIC method is implemented to measure the 3D full-field displacement and strain of complex dynamic deformations.

Finally, a DIC-based measurement system based on the above-mentioned research theory and methods has been developed for measuring complex dynamic deformations. A series of verification experiments have verified the feasibility, stability, and accuracy of the proposed measurement system. The research work has realized the 3D full-field measurement of some complex dynamic deformations, which has theoretical significance and engineering application value for measuring the large displacement deformation and the discontinuous deformation. It will expand the application scope of DIC.