Sistema embarcado de baixo custo projetado para dimensionar ovos de galinha através de visão computacional

Abstract

The use of new technologies for process automation in the Brazilian agro-industrial sector has been widely discussed in recent years, especially computer vision. In laying poultry, automation combined with digital image processing produces effective techniques to increase the quality of chicken egg production, ensure agility and reduce costs. In this work, a low-cost embedded system was developed for sizing chicken eggs using computer vision techniques. A prototype capable of providing an ideal environment - free from external noise - was designed to capture the photographs used in digital image processing, with autonomy for 4 hours of work without power interruption. The computer vision algorithm was developed using the Python programming language and the OpenCV open source library. The user has the option of interacting directly with the embedded system through a touchscreen display or accessing it remotely. Likewise, the results can be collected locally in the application or sent to another platform through the Bluetooth communication system, developed as a functional requirement for the construction of the system. The success rate for the measurements obtained through the system was 91% for the transverse axis and 89% for the longitudinal axis of the egg.