Automated visual inspection for metal parts based on morphology and fuzzy rules

摘要

Automated visual inspection system (AVIS) is a method of analyzing, classifying, detection defects for products at the production line. Usually, this inspection is either conducted by human, machine or both. In this paper, we explain an algorithm that capable to classify mechanical products in real time. The system is consists of two parts: hardware and software. The algorithm used the web-camera attaching to an adjustable arm to capture various image. Our main objective is to develop an image processing algorithm and fuzzy reasoning that can compute both the area and circularity of mechanical shapes and hence classify them according to their categories. The result shows the accuracy of classification is 80.5 % for group classification and 98% for individual classification of mechanical parts.