Ripeness evaluation of ‘Sun Bright’ tomato using optical absorption and scattering properties

• The hyperspectral imaging-based spatially-resolved technology was used to measure optical property of tomato.
• The relationship of optical property with ripeness of tomato was evaluated.
• The PLSDA achieved better classification accuracy for tomato ripeness.

Ripeness is one of the most important factors in determining the processing and eating quality of tomato. The objective of this research was to evaluate the changes of optical absorption and scattering properties in tomatoes during ripening, and develop classification models for grading the ripeness of tomatoes using optical absorption and scattering spectra. Optical absorption (μa) and reduced scattering (μs′) coefficients over the spectral region between 500 nm and 950 nm were measured, using a hyperspectral imaging-based spatially-resolved instrument, for 281 ‘Sun Bright’ tomatoes harvested at six ripeness grades (i.e., ‘Green’, ‘Breaker’, ‘Turning’, ‘Pink’, ‘Light-red’, ‘Red’). Absorption peak around 675 nm decreased consistently with the progression of ripeness, and its mean value for ‘Red’ tomatoes was almost zero. The reduced scattering spectra, on the other hand, decreased monotonically with the increasing wavelength; the value of μs′ also decreased successively from ‘Green’ to ‘Turning’, and an opposite trend was observed from ‘Pink’ to ‘Red’. Partial least squares discriminant analysis (PLS-DA) models yielded 92.1%, 84.4%, 92.3%, and 92.1% classification accuracies for the three ripeness grades (i.e., ‘Green/Breaker’, ‘Turning/Pink’, and ‘Light-red/Red’), when using the full spectra (500–950 nm) of μa, μs′, μa&μs′ and the effective attenuation coefficient (μeff = [3μa(μa + μs′)]1/2), respectively. The PLS-DA model using μeff achieved an overall classification accuracy of 88.4% for the six ripeness grades, which was 2.5% and 10.2% better than using the optical parameters of μa and μs′ alone. The research demonstrated that the optical absorption and scattering spectra, especially their combinations, are effective for classification of tomato ripeness.