Computare Vision Berbasis Camera dan Mini PC untuk Identifikasi Kecacatan Penutup Kemasan Minuman Kaleng

Abstract

Visual inspection becomes an important role when determining the quality of the conditions of production. Manual visual inspection carried out by relying on human senses, especially the eyes, the weakness is perspective and human's perspective may be different from each other, the quality result can be different. Utilization of the camera as a sensor and computer system as a major component of decision-making is the latest technology that is used as an automatic visual inspection replace human tasks. This research aims to make a visual inspection system that utilizes a web camera and mini PC (personal mini computare) Raspberry PI to identify disability packaging beverage cans. The system that is made has been tested to identify defects tin side of the inner surface of the cover packaging of canned drinks without color or silver color. The result is, the system is able to distinguish defective cans and the cans that are not defective with an accuracy of 66, 67%. We hope this research can be developed at a later stage so that automatic a visual inspection can be realized.