Identification of Crack Shapes by Digital Image Correlation Using Joint Estimation Maximum a Posteriori Method

Abstract A method that estimates invisible cracks from the surface based on deformation measured by digital image correlation (DIC) is being developed. An inverse problem setup to estimate such deformation. Surface deformation, DIC method, contains noise. Inverse problems have ill-conditions. The regularization applied in this study an extension of joint estimation maximum a posteriori (JE-MAP) method. JE-MAP algorithm alternates between MAP and grab-cut (GC) avoid physical constraints displacement forces at crack perimeters (ligaments) are added load ligaments cross-sparse relationship. or ligaments. estimated result varies greatly boundary This determined GC result. amplified changes results. results were input into improve boundary-determination accuracy. developed was combined with locations. proposed more accurately than L1-norm where observed data strain distributions