Concentration imaging of chlorophyll, carotenoids and anthocyanins in leaves

CORDON, G. B.; LAGORIO, M. G.

Mendoza

Congreso; 21st Interamerican Photochemical Society Conference; 2011

Inter-american Photochemical Society

A computational vision-methodology to determine the pigments concentration in plant leaves with spatial resolution from a digital image has been developed. For this purpose, reflectance spectra of uniformly colored leaves were recorded as a first step. Reflectance spectra were obtained from 300 to 700 nm, using a spectrophotometer equipped with an integrating sphere. Barium sulphate was used as 100% reflectance (white reference). Seven reflectance spectra were averaged for each species and the resultant spectrum was used for further derivation of colour coordinates (R, G and B). The selected species were: Hedera helix, Liquidambar styraciflua, Populus alba, Rosa sp., Gardenia jasminoides, Schefflera arboricola, Aloysia triphylla and Ficus benjamina. On these leaves, chlorophyll-a, chlorophyll-b, carotenoids and anthocyanins were quantified as usual by solvent extraction and subsequent spectrophotometric measurement. Values for the RGB coordinates were calculated from reflectance spectra, and they were correlated with the pigment concentrations. Values for the R component decreased non-linearly as chlorophyll concentration declined. Values for G component decreased linearly with chlorophyll content. Values for the B component, on the other hand resulted insensitive to chlorophyll concentration. An excellent correlation was finally found for the sum of R and G with the chlorophyll content. Other good correlations were found between the ratio R/G and the proportion anthocyanins/total chlorophyll and the ratio B/G and the carotenoids content. On the other hand, images from another set of colored leaves were captured using a commercial scanner (HP-DeskJet F380 de Hewlett-Packard) and they were digitally processed to obtain R, G and B bands. The scanned images in TIFF format (Tagged Image File Format) with a resolution of 300 ppp and a depth of 24 bits were digitally processed using the ERDAS IMAGINE 8.4 program. Using this software the TIFF images were imported to IMG format and they were subsequently separated in R, G and B bands. Using the correlations between the color coordinates (R, G and B) and pigments concentration, inferred from the homogeneous leaves, the concentration of pigments per pixel in the other set of leaves were estimated. The proposed methodology allows quantifying chlorophyll-a, chlorophyll-b, carotenoids and anthocyanins in a rapid and non invasive way, thus avoiding destructive solvent extractions, which requires too much work and time in lab. Additionally, image analysis shows a great potential due to the low cost of implementation, as it merely requires a commercial scanner and an image processing program.