High resolution topography utilizing a line scan stereo vision technique

Abstract

High-resolution surface topography measurement using two line scan CCDs was proposed for full-field optical inspection of surface geometry and defects. Based on traditional stereo vision technique, a new configuration of dual linear CCD system was constructed and a novel algorithm for correspondence matching based on digital image correlation (DIC) technique was proposed for the determination of 3D coordinates. Depending on the desired resolution, a user-defined template was employed in the algorithm to accurately select the correspondence pairs in the line scan image. The OpenGL™ database was used to reconstruct the surface topography from the corresponding matched points. Experimental results show that the accuracy and repeatability in the sub-micrometer scale were established with the proposed technique, which can be further improved depending on the pixel resolution of the line CCD. Since larger specimen can be inspected at high-speed by the use of larger arrays with smaller pixel size, the dual vision system can be easily used for high-resolution 3D profile measurement and defect inspection in-line.