The objective of this research was to investigate a multi-spectral scattering technique for measuring peach fruit firmness. A multi-spectral imaging system, which is capable of acquiring four spectral images simultaneously, was used to measure spectral scattering from “Red Haven” peaches, harvested at different ripening stages from two orchards, at wavelengths of 680, 880, 905 and 940?nm. Soft peaches had broader scattering profiles than firmer ones, which was most pronounced at 680?nm, the wavelength that is related to chlorophyll absorption. An empirical model, the Lorentzian distribution function with three parameters, was proposed to fit the restructured (shifted) scattering profiles for all four wavelengths, with the mean coefficient of determination (r²) equal to or greater than 0.998. Multi-linear regression models were developed on relating Lorentzian parameters to fruit firmness. The best firmness predictions (r²2 = 0.757 and the standard error for validation or SEV = 14.57 N or Newtons) were obtained when separate models were developed for each orchard. When data from two orchards were pooled, the model was able to predict fruit firmness with r² = 0.672 and SEV = 18.55 N. The multi-spectral scattering technique is non-destructive, fast and relatively easy to implement and it provides a non-destructive means for measuring peach fruit firmness.

Keywords: fruit, peaches, firmness, non-destructive sensing, multi-spectral imaging, scattering, Lorentzian function