Spectra from the on-line measurements were processed by another inhouse built MATLAB routine, separating potato spectra from conveyor belt spectra. NIR spectra from the conveyor belt differed from spectra obtained from potatoes. Separation of the two types of spectra could be based on intensity differences in the raw spectra, observed at four different wavelengths (i.e. the intensity of 800 nm had to be greater than the intensity of 760 nm and the intensity difference between 880 nm and 820 nm had to be greater than 350). Finally, an average of the three replicate measurements from each potato was calculated.