USO DE PROCESSAMENTO DIGITAL DE IMAGENS PARA O DESENVOLVIMENTO DE UM AMBIENTE AUTOMATIZADO PARA TESTES DE PLANTABILIDADE DE SEMENTES DE MILHO

• Ano de defesa: 2013
• Autor(a) principal: Dias, Wellington Cesar
• Orientador(a): Guimarães, Alaine Margarete
• Banca de defesa: Rocha, Jose Carlos Ferreira da , Fey, Emerson
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: UNIVERSIDADE ESTADUAL DE PONTA GROSSA
• Programa de Pós-Graduação: Programa de Pós Graduação Computação Aplicada
• Departamento: Computação para Tecnologias em Agricultura
• País: BR
• Palavras-chave em Português: milho; processamento digital de imagens; teste de plantabilidade
• Palavras-chave em Inglês: maize; digital image processing; plantability test
• Área do conhecimento CNPq: CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://tede2.uepg.br/jspui/handle/prefix/170

Nowadays, the process of determining the longitudinal distance between seeds in a plantability test is manual and prone to failures. For maize, the seed distribution in inadequate distances prevents each plant has the space needed for their growth, providing competition for nutrients and light , as well as plants arranged far apart, provide space for the development of weed. Considering the importance of testing plantability and that it has been done manually and prone to failures, this study aimed to develop an automated environment analysis of seed distribution, consisting of a combination of mat plantability, movie camera and software. A treadmill plantability metered seed and topped with carpet, which was incorporated into a frequency inverter to allow the mat to work at different speeds, the videos were recorded using the camera Contour+, and the software was developed in Java, implementing techniques of digital image processing appropriate to the problem at hand. The environment developed made possible to obtain an accurate reading of the distance between seeds by counting the elapsed time between each detection time of each seed crosses a certain video point and the speed of the treadmill. There were performed two different types of tests. In the test with previously known distances, the largest error obtained in the measured distances was 0.9 cm, which represents 2.25% at a distance of 40 cm from seeds, such as maize and the mean error was 0.19 cm which represents 0.48% at a distance of 40 cm from seeds. The tests with long measuring 500 seeds showed that the system detects the most seeds passing the mat, with the highest error rate 1.26%. The proposed environment also allows the detection of seeds with different colors, as in this research were used in natura seeds (yellow, extracted directly from a spike), reddish and bluish, according to the treatment received. It is believed that the solution can assist in the selection of faster and more efficient dosing discs along the plantability tests and consequently in obtaining a better arrangement of plants in the field, leading to higher productivity.