Research On Key Technologies And Applications Of Three-dimensional Digital Image Correlation

Three-dimensional digital image correlation (3D-DIC) is an efficient method for the full field non-contact optical measurement. Combing the digital image correlation and computer vision theory, it can accurately measure the three-dimensional surface morphology and deformation of the specimen. Compared with other optical measurement method based on interference, such as speckle interferometry, moire method and holographic interferometry, it has many advantages, such as the white light source, without vibration isolation system, wide measurement range, etc.This paper studies the four key technologies of the3D-DIC method:image acquisition, camera calibration, image matching,3D strain calculation. A newly developed acquisition hardware system is proposed. The camera calibration technology is described in detail. Firstly, the geometric model pinhole camera is provided, which link the physical location of surface point to the pixel position on the image plane. Then, the theoretical foundation for calibration is illustrated through several implementation steps, including (1) the initial estimation for single camera parameters;(2) single-camera parameters optimization;(3) dual cameras calibration. At last, to evaluate the surface strain, a local coordinate system must be established. After the displacement is fitted with a quadratic function, the strain tensor can be calculated by computing the function derivatives.The image matching technology is discussed from several aspects, including:(1) the random speckle pattern of subset matching;(2) the mathematical form of the correlation functions and their calculation characteristics;(3) subset shape function concept and its zero-order, first-order and second-order form;(4) sub-pixel matching algorithm, including optimization methods (Gauss-Newton algorithm and the Levenberg-Marquardt algorithm), matching strategy (forward additive, forward compositional and inverse compositional) and convergence criteria;(5) the initial estimation and improved full field matching strategy;(6) the image interpolation method;(7) the simulated speckle method for matching error assessment;(8) the matching method for the boundary subset;(9) the incremental digital image correlation method for large deformation matching;(10) The derivation of a newly proposed inverse compositional Gauss-Newton algorithm with the second-order shape function (IC-GN2), after simulation and experimental validation, this method is proved to be a high-efficiency and high-accuracy algorithm.Several real experiments are carried out to further verify the practicability of3D-DIC in mechanics application. The3D-DIC method is firstly applied to investigate the Portevin-Le Chatelier effect in Al alloy. The geometrical shape and propagation behavior of PLC bands are successfully obtained after analyzing the3D-DIC data. Then, six standard tests are performed to describe two data analysis method:cylindrical coordinate transformation and removal of rigid body motion.