Accurate 3-D Surface Profilometry Using Novel Boundary Edge Detection on Digital Image Correlation

Abstract

A novel 3-D edge detection methodology is developed to resolve the edge ambiguity problem encountered in 3-D optical surface profilometry employing digital image correlation (DIC). DIC has been surged as a full-field measurement technique for in-plane and out-of-plane dynamic mechanical structure analyses. However, up to date, one of the key issues in DIC is still remained in boundary edge detection since a surface edge is not detectable between two discrete neighboring height jumps due to optical diffraction. Generally, it is common to observe undesirable noisy measured data along surface edges in traditional DIC-based surface 3-D profilometry. To resolve this, a novel random speckle images processing method is established by proposing a new algorithm by employing the multiple symmetric partial template model to determine best edge location with accurate height reconstruction. A theoretical simulation on a pre-calibrated circle target was performed to verify the feasibility of the methodology. Some experiments on real industrial objects having various surface reflective characteristics were implemented to verify its capability on accurate detection of industrial objects having discrete surface edges. From its preliminary evaluation on measurement accuracy, it is found that the maximum measured error on critical dimension can be controlled within less than 6.0% of the overall measuring range while one standard deviation can be kept within less than 1.2%.