Diseño e implementación de un sistema de visión artificial para el control de calidad en etiquetado de productos comerciales

Abstract

The objective of this work was to develop an autonomous classifier algorithm that can identify whether a label passes quality control. For which, the design of a prototype of a conveyor belt was presented, which was later built-in steel plates and launched with the implementation of a DC electric motor whose torque is sufficient to move the conveyor belt and the product to analyzed, in the same way, the prototype has an arm that acts as a classifier and separates the products considered as accepted or rejected. The classifier algorithm was written in Python language in order to avoid costs related to payment of software licenses, it includes several libraries and resources that favored obtaining an image through a high-resolution webcam (720p/30fps) and processing of the same to carry out the respective comparison with a label with acceptable quality. In the same language, a user interface was designed and programmed where the classification process is shown in a simple and visual way so that the operator understands and takes control of the process that the algorithm executes internally. After the implementation of the prototype, together with the classifying algorithm, recognition calibrations were carried out, so that the algorithm provides reliable results to achieve the objective of speeding up and automating quality control in commercial product labeling. In this way, a functional algorithm was obtained that automates the repetitive work of controlling the quality of the final product. It is recommended that the algorithm programmer always be present for the initial start-up.