Banca de DEFESA: JHIORRANNI FREITAS SOUZA

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : JHIORRANNI FREITAS SOUZA
DATE: 25/08/2022
TIME: 14:30
LOCAL: Google meet
TITLE:

RELATIONSHIP BETWEEN SPECTRAL INDEXES OBTAINED BY DIGITAL IMAGES AND QUALITY ATTRIBUTES OF TIFTON 85 (Cynodon spp.)

KEY WORDS:

Principal component analysis; optical sensors; foragers; bromatological attributes.

PAGES: 102
BIG AREA: Ciências Agrárias
AREA: Engenharia Agrícola
SUBÁREA: Máquinas e Implementos Agrícolas
SUMMARY:

The world trade in fodder is expanding due to the demand for roughage to feed animals. The application of technology from computer vision can be an alternative to traditional methods of analysis of bromatological parameters, making the process less expensive and allowing to obtain non-destructive samples with results instantly. The present work aims to evaluate quality attributes of Tifton 85 grass (Cynodon spp.) through RGB and RGNIR digital images, relating the colorimetric indices with the bromatological analyzes of the culture. The experiment was carried out in an agricultural production area located at the Federal Rural University of Rio de Janeiro, in the municipality of Seropédica, RJ. The colorimetric indices obtained through digital images (in the visible and near infrared spectral region) of the cultivar Tifton 85 (Cynodon spp.) at different cutting ages (14, 28, 42 and 56 days) and fertilization levels (0 , 100, 200 and 300 kg/ha). ACP multivariate analysis was applied to extract the most relevant spectral indices and correlated with the original variables, through the criterion of PCs with explanatory power of at least 70% of the total variability of the data. PC1 values generated from the coefficients associated with the Colorimetric indices with the highest correlation were related by means of a simple quadratic regression with the nutritional attributes, to evaluate the productivity of dry matter, mineral matter, crude protein, ether extract, neutral detergent fiber, non-fibrous carbohydrate and gross energy. The relationship between the spectral indices and the quality attributes of Tifton 85 grass allowed the explanatory power of the variance in PC1 in almost all cases evaluated . The RGB spectral indices were more related to crude protein (R2 between 0.61 and 0.94), productivity (R2 between 0.61 and 0.94). The RGNIR spectral indices showed a greater relationship with productivity with R2 between 0.78 and 0.99, and with crude protein with R2 between 0.75 and 0.88. The RGNIR indices, in general, allowed a greater relationship with the bromatological parameters than the RGB indices.

 

BANKING MEMBERS:
Presidente - 2161955 - ANDERSON GOMIDE COSTA
Externo ao Programa - 3058101 - ANDRE MORAIS MOURA - UFRRJExterna à Instituição - AMELIA LAÍSY DO NASCIMENTO - UFRPE