Estimativa do índice relativo de clorofila em folhas do cafeeiro por meio de imagens digitais e ferramentas quimiométricas

Abstract

Coffee is among the crops of greatest economic relevance in Brazil. The high nutritional demand of coffee plants, particularly regarding nitrogen, makes continuous monitoring of plant nutritional status essential, since deficiencies directly impact yield and bean quality. In this context, chlorophyll stands out as an important indicator of leaf health and nitrogen use efficiency. Conventional evaluation methods, however, may involve high costs or operational limitations under field conditions. Therefore, the use of digital images and colorimetric parameters has gained prominence in precision agriculture. This study aimed to develop and validate models for predicting the relative chlorophyll index (RCI) in coffee leaves (Coffea arabica L.) using digital images acquired with smartphones. The images were obtained under real field conditions, with and without flash, using different smartphone models. As reference values, RCI measurements were performed using a commercial chlorophyll meter. From the images, colorimetric descriptors and indices derived from RGB channels were extracted and used to build chemometric models based on multiple linear regression (MLR), partial least squares regression (PLS), and least squares support vector machines (LS-SVM). The results indicated superior performance of the LS-SVM models compared to linear models, with higher coefficients of determination (R²) and lower root mean square error (RMSE) values in both calibration and validation. The best performance for set was observed for chlorophyll b using flash, with R² = 0.827, RMSE = 5.374, and r²m = 0.691 for the test set. The MLR models showed inferior performance, highlighting limitations in modeling the relationships between image descriptors and RCI. The proposed methodology offers a promising, accessible, and non-destructive alternative for nutritional management of coffee plants, with potential application under field conditions and benefits for small-scale producers. Keywords: Digital images; Plant leaves; Chlorophyll; Chlorophyllmeter; Chemometrics