Processamento de imagens digitais para detecção e quantificação de defeitos na madeira serrada de coníferas de reflorestamento de uso não estrutural
Carregando...
Arquivos
Data
2010-05-28
Autores
Rall, Ricardo [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
No Brasil, a utilização de Pinus (de reflorestamentos) na indústria madeireira tem sido crescente nos últimos anos. Particularmente para a madeira serrada desse gênero, a produção nacional de madeira serrada de Pinus atingiu 9,46 milhões de m³ em 2008. O crescimento da produção no período de 1999 a 2008 foi de 40,6%. A produção de madeira em ciclos mais curtos tem tornado comum o decréscimo constante na oferta de árvores adultas com grandes diâmetros. Além do aspecto da juvenilidade, nota-se que a madeira hoje disponível para uso industrial e comercial apresenta uma incidência muito grande de nós e outros defeitos, que comprometem significativamente seu desempenho físico mecânico e sua qualidade. Diante dos números que apontam uma crescente demanda por produtos de madeira serrada de Pinus, é natural que a indústria busque formas mais eficientes de garantir a qualidade dos produtos por ela gerados; neste sentido a atividade de inspeção assume um papel importante no processo produtivo. De fato, a classificação de madeira serrada é afetada sensivelmente pela qualidade do trabalho desempenhado pelos graduadores humanos (trabalhadores das serrarias), pois essa é uma atividade estressante, exigindo grande atenção por longos períodos de tempo. O objetivo geral foi o desenvolvimento de técnicas de processamento de imagens digitais para detecção de defeitos na madeira serrada de coníferas. Os objetivos específicos foram, por meio da análise digital das imagens, analisar quais os parâmetros devem ser empregados para a detecção de defeitos na madeira serrada de coníferas, além de desenvolver um software de visão de máquina para a detecção de defeitos na madeira serrada de coníferas. O material utilizado para o estudo foi a madeira extraída de seis árvores de Pinus taeda com 37 anos de idade, provenientes do Horto Florestal de Manduri – SP. As árvores...
In Brazil, the use of Pinus (reforestation) in the timber industry has been growing in recent years. Estimates indicate that the volume of lumber produced in the country, estimated at about 22 million m3, more than 35% is made of pinewood. Timber production in shorter cycles has become common for the steady decline in the supply of mature trees with large diameters (ABRAF, 2009). In the aspect of juvenility, it is noted that the wood available today for industrial and commercial use has a very high number of knots and other defects that significantly impair their physical performance and mechanical quality. Given the numbers that indicate a growing demand for sawn timber products from pine, it is natural that the industry seek more efficient ways to ensure the quality of products generated by it, meaning that the inspection activity plays an important role in the production process. In fact, the classification of lumber is affected significantly by the quality of work performed by human graders (sawmill workers), because this is a stressful activity, requiring close attention for long periods of time. So, the aim of this study was to develop techniques of digital imaging to detect defects in lumber conifers. The specific objectives were, through the analysis of digital images, analyze which parameters should be taken to detect defects in lumber conifers, and develop a software for machine vision detection of defects in lumber conifers. The material used for the study was the wood from six trees of Pinus taeda with 37 years of age, from the forest nursery of Manduri - SP. The trees were split into planks, which led to 84 samples in which they were originally analyzed manually nodes, presence of bone marrow and 17 other conditions that compromise their quality in accordance with standards established by the NBR 12297 (ABNT, 1991). In the processing of digital images acquired from these boards...(Complete abstract click electronic access below)
In Brazil, the use of Pinus (reforestation) in the timber industry has been growing in recent years. Estimates indicate that the volume of lumber produced in the country, estimated at about 22 million m3, more than 35% is made of pinewood. Timber production in shorter cycles has become common for the steady decline in the supply of mature trees with large diameters (ABRAF, 2009). In the aspect of juvenility, it is noted that the wood available today for industrial and commercial use has a very high number of knots and other defects that significantly impair their physical performance and mechanical quality. Given the numbers that indicate a growing demand for sawn timber products from pine, it is natural that the industry seek more efficient ways to ensure the quality of products generated by it, meaning that the inspection activity plays an important role in the production process. In fact, the classification of lumber is affected significantly by the quality of work performed by human graders (sawmill workers), because this is a stressful activity, requiring close attention for long periods of time. So, the aim of this study was to develop techniques of digital imaging to detect defects in lumber conifers. The specific objectives were, through the analysis of digital images, analyze which parameters should be taken to detect defects in lumber conifers, and develop a software for machine vision detection of defects in lumber conifers. The material used for the study was the wood from six trees of Pinus taeda with 37 years of age, from the forest nursery of Manduri - SP. The trees were split into planks, which led to 84 samples in which they were originally analyzed manually nodes, presence of bone marrow and 17 other conditions that compromise their quality in accordance with standards established by the NBR 12297 (ABNT, 1991). In the processing of digital images acquired from these boards...(Complete abstract click electronic access below)
Descrição
Palavras-chave
Madeira serrada, Processamento de imagens, Madeira, Pinus taeda, Wood, Pinus taeda L, Classification, Quality, Digital image processing, Sawn wood, Lumber
Como citar
RALL, Ricardo. Processamento de imagens digitais para detecção e quantificação de defeitos na madeira serrada de coníferas de reflorestamento de uso não estrutural. 2010. xiii, 123 f. Tese (doutorado) - Universidade Estadual Paulista, Faculdade de Ciências Agronômicas de Botucatu, 2010.