Banana (Musa sp.) maturity prediction system based on chlorophyll content using visible-NIR imaging

The chlorophyll content is one of the parameters to predict the maturity level of banana fruit. Nevertheless, the measurement of the chlorophyll is commonly destructive and performed manually through the biological test. In this paper, a banana maturity prediction system was introduced using Visible-Near Infrared (V-NIR) imaging based on the chlorophyll characteristic to estimate the maturity and the chlorophyll content non-destructively. The hardware of the measurement system consists of a set of sliders including controllable motor, Teflon table, halogen light source and a hyperspectral camera that connected directly to PC through Camera Link. The hypercube processing algorithms consist of reflectance image profile computation, spatial segmentation, spectral feature extraction, feature reduction, regression, and classification algorithm. The reflectance of the current image of the banana surface was corrected by the intensity value of the white and dark image. The spectral feature sets were computed using a principal component analysis on the full wavelength range of the camera spectra. The chlorophyll content was estimated using principal component regression. Thus, the maturity stage of banana was classified using support vector machine into three classes i.e. immature, mature and very mature based on the chlorophyll profile characteristic. The proposed system was evaluated using 45 Ambon bananas (Musa acuminata colla) samples which consist of 15 samples for each maturity stage. The correlation coefficient is 0.89 and RMSE value is 5.98 × 10^-4 %. The maturity classification error using five folding of cross-validation is 2.1%. The results show that the proposed system can predict the banana maturity stage perfectly and suitable in an industrial sorting system for banana fruit quality.