Non-destructive Estimation of Chlorophyll a Content in Red Delicious Apple Cultivar Based on Spectral and Color Data

Abstract

Non-destructive estimation of the chemical properties of fruit is animportant goal of researchers in the food industry, since onlineoperations, such as fruit packaging based on the amount of differentchemical properties and determining different stages of handling,are done based on these estimations. In this study, chlorophyll acontent in Red Delicious apple cultivar is predicted as a chemicalproperty that is altered by apple ripening stage, using nondestructive spectral and color methods combined. Two artificialintelligence methods based on hybrid Multilayer Perceptron NeuralNetwork - Artificial Bee Colony Algorithm (ANN-ABC) and Partialleast squares regression (PLSR) were used in order to obtain a nondestructive estimation of chlorophyll a content. In application of thePLSR method, various pre-processing algorithms were used. Inorder to statistically properly validate the hybrid ANN-ABCpredictive method, 20 runs were performed. Results showed that thebest regression coefficient of the PLSR method in predictingchlorophyll a content using spectral data alone was 0.918. At thesame time, the average determination coefficient over 20 repetitionsin hybrid ANN-ABC in the estimation of chlorophyll a content,using spectral data and color features were higher than 0.92±0.040and 0.89±0.045, respectively, which to our knowledge is aremarkable non-intrusive estimation result.