Predictions of chlorophyll concentrations in the leaves of seedlings of two congeneric tropical trees from RGB digital image components

Abstract

The segmentation of digital images in red, green and blue (RGB) components is a low-cost method for monitoring leaf chlorophyll concentrations and seedling quality. The two congeneric species, Cariniana legalis and C. estrellensis, are distinguished based on differences in bark texture and the colour of their new leaves. We compared indices based on leaf colour segmentation in RGB to predict total chlorophyll concentrations (Chlt) in the leaves of seedlings of these two species. Mature leaves were digitalised in a flatbed scanner and segmented in red (R), green (G) and blue (B). The relationships between the three RGB indices and Chlt were tested. Additionally, we calculated the anthocyanin content-chroma basic (ACcb). The mean value of ACcb was significantly higher in C. legalis than in C. estrellensis, demonstrating a higher anthocyanin concentration in C. legalis leaves. Based on the highest coefficients of determination (R²) and lowest prediction errors (PE), for all indices, the best results were obtained for C. estrellensis. The presence of anthocyanins in the leaves of C. legalis and the limitation of the RGB colour segmentation indices for separating all leaf pigments might be the main causes of the differences in Chlt prediction in the leaves of these two congeneric tree species.