Establishment of method distinguishing between defects and artificial texture in defect inspection with phase-shift illumination method

Abstract

Inspection systems with a machine vision camera have been extensively applied in factory automation detecting unclear defects. In this automatic inspection, optical engineering technology needs to emphasize the contrast of defects in order to quickly find them in a camera image and improve image recognition accuracy. Using the phase-shift illumination method with striped structured illumination, we develop optical engineering and image processing technology to enhance defects in transparent materials. Our challenge has been to distinguish actual defects from dark fringes due to artificial three-dimensional texture on a target sample. We proposed theoretically that an innovative method providing an illumination pattern with finite-width dark regions would make the gentler slope of artificial texture less visible than actual defects. We extended our theoretical model in the case of a square wave as the typical illumination pattern and established an inspection method distinguishing defects from artificial texture. We confirmed with simulations and experiments that the square-wave illumination enables us to distinguish between defects and artificial texture.