Plugin Seeds Analysys® na avaliação morfocolorimétrica de plântulas de gergelim

• Ano de defesa: 2024
• Autor(a) principal: Oliveira, Markson Luan do Vale
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Visão computacional; Morfometria; Colorimetria; Vigor; Sesamum indicum; Sementes; Computer vision; Morphometry; Colorimetry; CNPQ::CIENCIAS AGRARIAS::AGRONOMIA::FITOTECNIA
• Link de acesso: http://repositorio.ufc.br/handle/riufc/79086

Sesame is an oilseed with a high oil content, propagated by seeds, and the quality of these seeds is crucial to increase productivity. Seed quality is assessed by physical, physiological, genetic and health tests; these methods are time-consuming, and can be subjective and inaccurate. In this context, computer vision proposes faster and more accurate alternatives for seed evaluation. The objective of this study was to verify the efficiency of the ImageJ Seeds Analysis® plugin in the morphocolorimetric evaluation of sesame seedlings and in distinguishing the quality of the lots. The research was conducted at the Seed Analysis Laboratory, of the Department of Plant Science, at the Center for Agricultural Sciences of the Federal University of Ceará, Fortaleza, CE, in collaboration with the University of Cagliari, Italy. The methodology was divided into two stages. In the first stage, 15 sesame lots were analyzed using traditional tests: first count, germination, shoot and root length, shoot and root dry mass weight, emergence, emergence speed index, electrical conductivity, and 1000-seed weight. In the second stage, the germination test was performed to capture images of seedlings at the end of the test with a flatbed scanner. These images were analyzed using the Seeds Analysis® plugin of ImageJ to extract morphocolorimetric features (area, perimeter, feret, rfactor, solidity, and means of red, blue, green, RGB, gray, H, S, and V). Statistical analyses included Tukey's mean tests, Spearman's correlation, and principal component analysis. The results showed significant variations in seed vigor, with high correlation between the variables first count, germination, percentage and emergence speed index and the Mean RGB and H features. The correlation was also high between first count and germination with mean blue, and between shoot length and solidity and mean S. It is noteworthy that mean red had a high correlation with several variables (first count, germination, shoot and root dry mass, percentage and emergence speed index). Principal component analysis allowed grouping of lots based on their characteristics. The attributes rfactor, breadth, perimeter, area, aspratio and feret formed a group with lots 11, 15, 9, 7, 10, 3, 12, 8, 2 and 4. The attribute solidity grouped lots 5, 6, 13 and 7. Lots 3, 4, 6, 7, 8, 12, 13, 14 and 15 were associated with mean green, blue, H, RGB and gray, while lots 10, 1, 2, 8 and 9 were grouped with mean S, red and V. These groupings indicate correlations between the morphocolorimetric attributes and the quality of the lots. It is concluded that the Seeds Analysis® plugin was effective in extracting morphocolorimetric features, and that morphocolorimetric analysis is an effective and reproducible tool for evaluating the vigor of sesame seed lots and can be incorporated into traditional seed evaluations. Spearman and principal component analyses were important for highlighting the morphocolorimetric features that were related to and most influenced the variables of traditional tests.