Aplicação de processamento de imagens para medição do nível do leite na coleta a granel

Abstract

The present work exhibits the implementation of a prototype that performs the reading of the ruler in the measurement of the level of milk in the bulk collection through image processing, offering an alternative to the manual reading, due to the fact that the step of measuring the level of milk in bulk collection is subjective and subject to human errors, which can cause damage to the producer and the dairy. A portable photo studio was developed, with which some pictures were taken. The algorithm was tested with these photos and was successful in determining the level of milk in the tank in all the photos collected, obtaining mean absolute error of 0.086cm and mean absolute percentage error of 0.255%, demonstrating the feasibility of the proposed technique, obtaining low errors, small processing time and reaching the objectives.