AI-based Restoration Reflection Region Method for 3D Automatic Optical Inspection

Abstract

This study focuses on a three-dimensional measurement system of components on a printed-circuit-board (PCB). Most existing methods for automatic optical inspection face the problem of light reflecting on components. We herein propose a new method to restore the reflection area of components with the aim of measuring their shape accurately. This method use a pattern projection system along with an algorithm for restoring the components’ reflection area. For efficient restoration of the restore reflection area, the reflection regions are first identified through pixel resolution, wherein, saturated pixels represent the reflection regions. Then, the PCB components are tilted slightly using solder. These characteristics are used to, obtain the rotation matrix of a component, and its reflection area is restored. The performance of the proposed method is evaluated through a series of experiments, whose. results demonstrate its effectiveness.