ABSTRACT
Automatic optical inspection is a widely explored technique that is used for surface flaw detection; however, most of the previous studies have focused on inspecting flaws on a plane; few studies are for spherical surface flaw inspection. This article explores such a spherical surface flaw inspection problem via the acquisition of whole spherical surface images, using four Charge-Coupled Device (CCD) cameras and a positioning mechanism, to adjust the view angles of the cameras. Twelve view angles of images along with the image stitching and fusion techniques are proposed to form a full spherical surface image. This study has built up a positioning mechanism for CCD image capture and implemented the techniques to achieve a full golf-ball surface image for several experimental cases. The experimental results show that the proposed method is feasible and useful for spherical surface flaw inspection. The machine adopts an automated inspection process to quickly capture images and shorten the lead time of the job. Using a PC computer with LabVIEW software analysis, the image coordinate conversion and fusion process provide an effective global image for the detection of the ball.
