Sistema computacional para integração de dados na análise de cafés especiais

• Ano de defesa: 2016
• Autor(a) principal: Leme, Dimas Samid
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Lavras; Programa de Pós-Graduação em Engenharia de Sistemas e Automação; UFLA; brasil; Departamento de Engenharia
• Palavras-chave em Português: Café - Torração; Análise sensorial; Sistema de visão computacional; Redes neurais artificiais; Banco de dados; Aplicação Android; Coffee - Roasting; Sensory analysis; Computer vision system; Artificial neural networks; Databases; Engenharias
• Link de acesso: http://repositorio.ufla.br/jspui/handle/1/11071

The shade of color of roasted coffee varies in light of the production objective. However, there is an international standard for the degree of roasting used in sensorial analyses, measured by means of a high costing equipment that, in some models, does not allow storing the results in a data integration system. The Computational View systems emerge as alternatives for a quick analysis, storage and integration with other data concerning coffee. Thus, the objective of this work is the construction of a computational view system for identifying the different shades of roasted and milled coffee grains. For this, a conversion of the RGB color standards of digital cameras was performed for parameters L*, a* and b* of each pixel of the digital image, obtaining an average of all pixels of the sample. For creating the computational view system a closed metallic structure, illumination system standardized by LEDs, a digital camera attached in its superior side and processing software of the images implemented with polynomial regression models and artificial neural networks for approximating a function that represents the most accurate roasting degree of the photographed samples were used. For constructing the transformation model of color spaces, a databank of color charts and 150 samples of roasted coffee in different shades for training an artificial neural network were used. With the results obtained, it was verified that the model presents good accuracy with low divergence. Furthermore, Android/iOS applications we developed for registering sac data, physical and sensorial analysis data defined by the American Association of Special Coffees (SCAA). These applications also allow taking the temperature of samples and posting to an integrated platform with low implementation cost if compared to other tools available.