Location: Tropical Crops and Germplasm ResearchTitle: Evaluating Hass Avocado Maturity Using Hyperspectral Imaging Author
Submitted to: Caribbean Food Crops Society Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 6/3/2008
Publication Date: 9/1/2008
Citation: Girod, D., Landry, J.A., Doyon, G., Osuna-Garcia, J.A., Salazar-Garcia, S., Goenaga-Portela, R.J. 2008. Evaluating Hass Avocado Maturity Using Hyperspectral Imaging. Caribbean Food Crops Society Proceedings. 44(2):144-154. Interpretive Summary: The maturity of avocado fruit is usually assessed by measuring its dry matter content(DM through a destructive and time consuming process. The field of spectroscopy is an emerging technology in the agro-food industry and is seeing new applications in the assessment of produce quality and safety. This study tested the use of hyperspectral imaging as a mean to estimate the DM content of avocado quickly and non-destructively. Visible/Near Infrared spectra of intact avocados were collected over the 400nm-l000nm range of the electromagnetic spectrum, with a 2.5nm resolution in reflectance mode, resulting in 163 spectral bands. Reasonably accurate models (R2>0.8) were obtained for DM content with the entire spectral range. This study demonstrated that hyperspectral imagery serves as a potential tool to evaluate the maturity of fresh avocado. There was a strong correlation between the dry matter content of Hass avocado, and its skin color (as perceive by spectroscopy).
Technical Abstract: The maturity of avocado fruit is usually assessed by measuring its dry matter content(DM), which is a destructive and time consuming process. The aim of this study is tointroduce a non-destructive and quick technique that can estimate the DM content of an avocado fruit. 'Hass' avocado fruits at different maturity stages and varying skin colorwere analyzed by hyperspectral imaging in reflectance and absorbance modes. The DM ranged from 19.8% to 42.5%. The hyperspectral data consist of mean spectra of avocados in the VISINIR region, from 400nm to 1000nm, for a total of 163 different spectral bands. Relationship between spectral wavelengths and DM content were carried out using a chemometric partial least squares (PLS) regression technique. Calibration and validation statistics, such as correlation coefficient (R2) and prediction error (RMSEP)were used as means of comparing the predictive accuracies of the different models. The results of PLS modeling, over several different randomizations of the database, with full cross validation methods using the entire spectral range, resulted in a mean R2 of 0.86 with a mean RMSEP of 2.45 in reflectance mode, and a mean R2 of 0.94 with a mean RMSEP of 1.59 for the absorbance mode. This indicates that reasonably accurate models(R2>0.8) could be obtained for DM content with the entire spectral range. Also this study shows that wavelengths reduction can be applied to the problem. Starting with 163 spectral bands, the DM could be predicted with identical performances using 10% of the initial wavelengths (16 spectral bands). Thus the study demonstrates the feasibility of using VISINIR hyperspectral imaging in absorbance mode in order to determine a physicochemical property, namely DM, of 'Hass' avocados in a non-destructive way. Furthermore it gives some clues about which spectral bands could be useful for that purpose.