Design of automated system for online inspection using the convolutional neural network (CNN) technique in the image processing approach

Abstract

—Presently, many achievements in various fields such as dyeing, textile or packaging industry have been significantly gained. In this situation, the large scale of products has produced without any efficient inspection methods. It requires workers to detect manually by supervising the hundreds of meters of fabric or nylon. To avoid inconveniences or mistakes by human in manual operation, a novel idea to develop the automated inspection system is mentioned. The main architecture of this system comprises computational mechanics of several components, equipment placement and installation. Owing to the heavy load, two servo motors and chain drive are suggested to integrate in the driving mechanism. In this research, the principal factors to identify defect on surface are advanced techniques of computer vision. Several filters and image processing methods are implemented while the motion of rolls is executing. To validate our works, the real-world platform of proposed approach is entirely fulfilled. Some tests have been applied in this hardware in order to obtain the practical results. From these achievements, it could be obviously proved that our approach is feasible, efficient, and applicable for related industries.