The Nondestructive Assessment of the Potatoes Internal Quality by Using the Method "Seeing Through Layers" (STL)
ABSTRACT It has been studied 628 potato tubers taken from different European regions.
The screening group of potatoes includes 20 varieties, every one of which is characterized by its own specific type of skin and flesh. The samples themselves are scanned in the process of gravitation falling during their passage through the measuring zone of photometric camera, developed for the assessment of potatoes being in flow (on-line).
Each sample (individual tuber) was scanned both unpeeled and physically peeled. During this process organoleptic evaluation was made of the appearance and internal quality of the sample.
On the basis of obtained data it has been optimized the mathematic statistical models for prediction of the internal optical density, by which models the virtual peeling is performed.
It has been made a classification (unpeeled, physically peeled and virtually peeled tubers) by using the K-nearest neighbors (KNN) method, by which it has been proven
the a priori, intuitional expectations:
- The spectrum of the samples skin is a disturbing factor and the elimination of its influence is going to increase the precision of potatoes sorting.
- By the method “Seeing through layers” it is carried out a more precise sorting of potatoes on the basis of the elimination of the skin disturbing effect.