IMPACTO DA ANÁLISE DA NITIDEZ EM MÉTODOS DE CLASSIFICAÇÃO DE IMAGENS DE MADEIRA

• Autor(a) principal: Barros, Luiz Gustavo
• Data de Publicação: 2013
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UEPG
• Texto Completo: http://tede2.uepg.br/jspui/handle/prefix/176
• topic: análise de nitidez; visão computacional; classificação de imagens; sharpness; computer vision; image classification; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO

Images acquired in the field with ordinary cameras, tend to have a lower quality compared with those obtained in the laboratory, including noise and unclear areas. This paper proposes a methodology for the analysis of sharpness in wood images of wood to analyze its impact on their classification. To evaluate this proposal, a database of wood images was used and each sample has extracted its color and texture characteristics using uniform local binary pattern and RGB color model methods, respectively. The Tenengrad method was used to evaluate the image sharpness and then the images were classified as high or low sharpness. Scenarios were developed to test the impact of the analysis of sharpness and check how it contributes to the hit rate.