Preliminary study of the relation between the content of cadmium and the hyperspectral signature of organic cocoa beans

Abstract

The contamination of soils by heavy metals is a current problem for agricultural production. Rapid access and reliability to heavy metal concentration such as cadmium is crucial for international trade. In the present study, visible and near infrared (VIS-NIR) spectroscopy, combined with linear and statistical methods, were used to predict the cadmium concentration of organic cocoa bean samples. Partial Least Square Regression (PLSR) and Support Vector Regression (SVR) were implemented to estimate the content of this heavy metal from hyperspectral imaging and chemical analysis. Competitive Adaptive Reweighted Sampling Method (CARS) and Jackknife method were used for selecting optimal wavelength. The SVR model performed satisfactorily with the use of 45 resulting wavelengths from optimization using CARS and the Jackknife method, with an adjusted coefficient for the test R2 of 0.9401 and an RMSEP of 0.2594. Based on the results, it was concluded that VIS-NIR spectroscopy combined with CARS-Jackknife methods seems to be a fast and effective alternative to classical methods for predicting the concentration of cadmium in organic cocoa beans.