Parallel reliability-guided algorithm for digital image correlation
Digital image correlation is a non-contact optical method for measuring the displacement and strain on the surface of a material. The existing reliability-guided digital image correlation (RG-DIC) method is stable and reliable for a single image but it still needs a large calculational resource for a sequence of images. Due to the decorrelation effect, the reference image must be replaced several times to correct the measurement results for an image sequence involving a large deformation or a discontinuous deformation. Since the process must be executed sequentially, image by image, the total time required is often unacceptably large when the image sequence is long. The challenge is to find a way of improving the speed while retaining calculational reliability and measurement accuracy, which are important for the practical application of DIC. To address this problem, an improved method is proposed in this paper. The parallel bottleneck caused by the decorrelation effect is solved through improving the parallelism to increase the processing speed. This approach can be used to calculate the strain field of the surface of the material in cases of discontinuous deformation, such as in the area near to a crack. Compared with existing methods, this method not only retains the calculational reliability but also greatly improves calculation speed, especially on current multi-core computing platforms.