Diseño, construcción y automatización de un prototipo de máquina clasificadora de piñas de acuerdo al color y forma por medio de visión artificial

Abstract

Automated processes have been gaining relevance due to their coupling to different types of products; in the case of fruits, studies have been developed to avoid direct contact in the classification process. In the present work, the design, construction, and automation of a prototype of a pineapple sorting machine were carried out according to the color and shape by means of artificial vision. For the design, the alternative analysis method was used to find the best solution that guarantees a process without contaminating or damaging the pineapple. The solution consists of the transport system by means of rollers, the acquisition of the image by means of a Raspberry Pi camera within a controlled lighting system, the classification system by means of the pneumatic cylinders, and the exit of the fruit by trays. SolidWorks and SAP2000 software was used for the modeling and verification of the mechanical elements. The algorithm was developed in Python by using OpenCV libraries. A segmentation based on thresholding was applied, for the extraction of characteristics by color the number of valid pixels was obtained, and for characteristics by shape, the morphological operation dilation and Hu Moments were used. The classification both by color and shape was carried out by comparing a value called the threshold. A graphical user interface was implemented using a touch screen and the prototype was controlled through the Raspberry Pi. For the results, a comparison of the manual classification and that of the prototype was made. Effectiveness of the prototype of 96.99% and a reduction of the time spent of 13.65% was obtained. It was concluded that the prototype performs the classification process without affecting the pineapple and it is recommended that processors speed up execution times.