On improving the accuracy of self-calibrated stereo digital image correlation system

Stereo digital image correlation (stereo DIC), a full-field deformation measurement technique, is increasingly being used to resolve strains at µm-length scale by using microscope-like imaging systems. Self calibration of these imaging systems is more cost-effective and convenient than the conventional target-based calibration. Though the use of self-calibrated stereo DIC systems has already been reported, less attention has been paid to improving the accuracy of these systems. In the present work, we improve the accuracy of a self-calibrated stereo DIC system, which is composed of two full-frame DSLR cameras coupled to macro lenses and is used for testing ASTM E8M sub-sized flat dog-bone specimens. First, we collect the images of two of the speckled test specimens that subtend an angle of 12 between them using an f/25 aperture. Our image-collection strategy leads to a convergent imaging configuration with viewpoints that range from -45 to 45 across two perpendicular directions. Next, we process the collected images in a commercial photogrammetric calibration software by using more than nine image points for computing each object point. We validate our findings on a rigid-body motion test and a uniaxial tensile experiment, and we observe an excellent agreement between the stereo-DIC measurements and the ground truth. Using our findings, the reprojection error of self calibration is improved from 0.3 pixel to 0.1 pixel. The error in the stereo-DIC strain measurements is always less than 3.4% with the improvements made to self calibration, whereas it is as large as 7.6% without them.