Firmness Prediction System of Waxed Malang Apples using Hyperspectral Imaging

Abstract

These days, the wax coating was applied on fruits to maintain its quality and extend the shelf life. The quality measurement of waxed fruits was destructive in most cases. In this study, hyperspectral imaging was used to predict the quality of the fruits non-destructively. The quality of fruits that predicted was firmness. Firmness is one of the parameters that determine the maturity of the fruits. The objects that used was Rome Beauty variety of Malang apples. Wax coating on Malang apples used wax emulsion made of beeswax, coconut oil, and sunflower oil. Image acquisition of Malang apples used reflectance mode with wavelengths 400-1000nm. Image processing steps included image correction, Region of Interest (ROI) selection, feature extraction, dimension reduction, and regression model. Partial Least Square Regression (PLSR) was used as a dimension reduction and regression model algorithm. The prediction model was built using non-waxed Malang apples, waxed Malang apples, and a combination of non-waxed Malang apples and waxed Malang apples. Root Mean Square Error (RMSE), Determination Coefficient (R2), and Residual Predictive Deviation (RPD) are evaluation parameters used to determine the performance of the model. The performance of model PLSR using waxed Malang apples were 0.96 for R2; 4.38 for RMSE, and 2.13 for RPD respectively. Based on these results, the firmness prediction system can be implemented to measure the quality of waxed fruit non-destructively.