However, it should be pointed out that three healthy kernels were ‘‘wrongly” identified as the contaminated using the threshold β = 0.15, which is out of accord with in-kind shooting pictures as in Figs. 3 and 4. The reasons may be as following: 1) contaminated pixels of the three ‘‘wrongly” identified kernels have the chance of being getting moldy; 2) slightly contaminated peanuts could be normal to human eyes, but they do have aberrant spectral response. This is exactly the strengths of SWIR hyperspectral imaging seeing objects at broader wavelength range far beyond visible. Furthermore, it is preferable to rigorously decline any contaminated-like kernels rather than to easily accept them. However, it must be pointed out that chances are that threshold b can inevitably bring in detecting false negatives, which is unacceptable because some moldy pixels can accepted as healthy. Again, it is suggested that the threshold β should be as small as possible.