Classifier / selector machine for oranges using computational vision

• Grant number: 03/08464-4
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Start date: July 01, 2004
• End date: August 31, 2007
• Field of knowledge: Agronomical Sciences - Agricultural Engineering - Agricultural Machinery and Implements
• Principal Investigator: Reinaldo Augusto da Costa Bianchi
• Grantee: Reinaldo Augusto da Costa Bianchi
• Company: Digital Vxia do Brasil Automação Ltda
• City: São Paulo

Abstract

The cultivation of oranges has experienced vertiginous growth in the last few years owing to the increased internal and external market. Among the factors that have driven this expansion is the devalued exchange rate, the policy of strengthening exports and standardization and quality control initiatives. One of these initiatives was the creation of the Brazilian Program for the Modernization of Horticulture, in which a norm for the classification of oranges was established. What has prevented many producers from adhering to the program is the fact that it employs human visual inspection, which is slow, expensive and often outside the expected norms. This project intends to develop an orange classifying/selecting machine suited to the Brazilian economic reality and which complies with the established standardization system. This technological innovation is opportune given that no similar machine is built in Brazil which can undertake the classification of oranges not just according to weight and size, but also according to their visual attributes; and that, in contrast to imported machines, is suited to the characteristics of the national fruit, offering local support and maintenance, in addition to a price accessible to the small to medium producer. With a prototype built and tested, which processes an orange per second, the aim of the second phase of the project is to evolve to a machine capable of processing around 15 oranges per second. (AU)