Abstract
Peeling is essential in potato processing, yet conventional assessment methods face issues like inefficiency and environmental concerns. This study proposed a hyperspectral imaging approach combined with dual-threshold segmentation to quantify potato peeling rates. This method was in the preliminary research stage. The objectives of this work were to experimentally validate the feasibility and accuracy of this method. A hyperspectral system captured images of potatoes after water-jet peeling. By analyzing spectral data from peel and flesh regions and applying principal component analysis, the key wavelength of 592 ± 20 nm was identified, where the reflectance difference between flesh and peel was most pronounced. Grayscale images derived from this band were processed via median filtering and dual-threshold segmentation to differentiate flesh from peel. The peeling rate was calculated as the pixel ratio of flesh to total potato area. In validation tests, the calculated peeling rates showed an average absolute error of 0.69% compared to manual measurement, confirming the potential and feasibility of this technique. The proposed method offers a promising, non-destructive, and eco-friendly alternative for monitoring peeling quality in agricultural processing, though further research is warranted.