Comparison of chemometrics methods for classification of sugarcane brix using visible and shortwave near-infrared technology

The potential of visible and shortwave near-infrared (VSWNIR) diffuse reflectance spectroscopy in the range of 400 to 1000 nm, in combination with three classifier algorithm techniques, was investigated to classify sugarcane quality into three quality classes (high, medium and low). Two hundred and ninety sugarcane internode samples were used to evaluate the ability of this technology. Each internode sample was scanned at four scanning points to obtain the spectral data which was later correlated with its sugar content (Brix value). The Brix values were later classified using three classifier algorithms namely Bayesian discriminant analysis (BDA), artificial neural network (ANN) and support vector machine (SVM). The results shows that the overall classification accuracies achieved by BDA, SVM and ANN were 77.8, 83.1 and 88.7% respectively. The results demonstrated that the VSWNIR spectroscopy together with chemometrics techniques could be a rapid tool to be used for classification of sugarcane quality based on spectral data