Image Based Surface Visualization Method for the Defects Detection

Abstract

Defects detection on the final product of camera module is a major issue in the visual inspection industry. This paper is about the study of surface visualization for foreign material detector or abnormal lens surface between IR filter and lens in camera module. With the purpose of making filter surface noticeable, it extracts background image assumed as polynomial expression in input image and then generate the differential image of input image. Also, it accentuates flexuosity of surface, applying median filtering with high precision, after it smoothes a set of pixels that has recursively similar brightness value based on coarse-to-fine strategy known as bilateral filtering. Performance of the proposed method has been evaluated on some sample images captured in the white light. As a result of the experiment, we verified that surface condition is visualized and surface defects such as foreign substance is identified distinctly. And this method is expected to be used widely in the machine vision industry for automated or manual visual inspection of defects system.