Non-destructive prediction of quality of intact banana using spectroscopy

Potential of spectroscopy in the wavelength range of 299–1100 nm for nondestructive prediction of dry matter (DM) content, pH, sweetness (total soluble solids, TSS), and acid–Brix ratio (ABR) for banana was studied to have proper estimation of their maturity and ripening stage. Partial least square (PLS) and multiple linear regression (MLR) employing pre-processing techniques were carried out. PLS models were found to be the best with multiple correlation coefficients for calibration and validation were found to be 0.88 and 0.81 °Brix for TSS, 0.85 and 0.78 for ABR, 0.88 and 0.83 for pH and 0.90 and 0.87% for DM, respectively. The standard errors of calibration, prediction, biases and differences in them were low, which indicated that NIRS has potential to predict maturity and ripening parameters of banana nondestructively.

► Calibration models (PLS and MLR) developed for different groups of wavelengths in the range of 299–1100 nm could predict TSS and DM fairly in the wavelength range of 955.235–981.879 and 1063.248–1089.407 nm, respectively without any data treatment and pH in the wavelength range of 927.927–954.690 nm after baseline correction.
► Treatment of data with MSC improves the prediction of ABR.
► The calibration models so developed will be helpful in designing of portable offline and/or online instruments for quick grading of fruit.