High‐speed, two‐dimensional digital image correlation algorithm using heterogeneous (CPU‐GPU) framework

Abstract

Two-dimensional digital image correlation (2D-DIC) is an experimental technique used to measure in-plane displacement of a test specimen. Real-time measurement of full-field displacement data is challenging due to enormous computational load of the algorithm. In order to improve the computational speed, the focus of recent research works has been on the approach of parallelization across subsets within image pairs using graphics processing unit (GPU). But alternate GPU-based parallelization approaches to improve the performance of this algorithm as per the order of data processing have not been explored.To address this research gap, our method utilizes parallelism within a subset as well as across subsets for each computation step in an iteration cycle. A heterogeneous (CPU-GPU) framework in combination with a pyramid-based initial values estimation for subsets (in parallel) is proposed in this work. The precompute steps of the proposed framework are implemented using CPU, whereas the main iterative steps are realized using GPU. It is demonstrated that the overall computational speed of the proposed heterogeneous framework improves by ≈ 9× compared to a sequential CPU-based implementation for a pair of gray-scale images with a resolution of 588 × 2, 048 pixels. As an important milestone, feasibility to measure deformations in real time (≤1 s) is manifested in this study.

Application of high‐performance DIC for a comprehensive evaluation of biaxial fatigue crack growth experiments

“…Both implementations use zero‐order shape functions, which are sufficient for deformations occurring in fatigue trials. For comparison, Thiagu et al [ 12 ] report a correlation rate of about 1.5 kHz for a GPU‐based DIC implementation with 21 × 21 pixel subset size, allowing for an iterative evaluation of higher‐order shape functions as they are required for more complex deformations—which, according to the authors, is still a speedup by a factor of 9 compared to CPU implementations.…”

“…[9][10][11] Real-time DIC data processing is challenging due to the enormous computational load of the related algorithms. Therefore, graphics processing unit (GPU)-based 2D and 3D implementations have become increasingly popular in recent years, [12] with almost all of them concentrating on full-field evaluations. Without GPU, even integral strain measurement rates correlating only two or four subsets hardly exceed 100 Hz.…”

Center Atom Model for Strain Mapping of Void and Crack of Atomic Lattice Image

“…[24] Along with the great progress of the computer digital-image processing technology, the point-to-point resolution recognition technology of the super microscope technology has also made a breakthrough improvement. [25][26][27] The images of high-resolution transmission electron microcopy (HRTEM) at nanoscale make it possible to accurately process local elastic strain images, [27] which are also used to characterize the elastic strain field images of the lattice distortions of the grain boundaries, dislocations, interfaces of nano-microelectronic devices, and hetero-junction interface structures. [24,28,29] In addition, there are some postprocessing methods based on atomic coordinates, such as calculating the atomic-level elastic strain tensors in crystalline systems on the elastic-plastic decomposition of crystal deformation, [30] calculating local strain tensors based on the relative motion of neighboring particles [31,32] and using the polyhedral template matching method to calculate strain.…”