IS*~*p*~ value is equal to 3. For the same number of different *E*~*j*~ values (*K* = 10), the empirical values for the *IS*~*p*~ and *IS*~*c*~ are both approximately one, which, as expected, is close to the theoretical value of unity. This is a surprising result, since the overlap between these 2 bands should be much higher than what *IS*~*p*~ and *IS*~*c*~ values suggest. However, while values for *IS*~*p*~ and *IS*~*c*~ are close to 1.0, the *KS*~*p*~ and *KS*~*c*~ values are at around 0.58 and 0.62, respectively. Finally, the total number of classes *K*~*T*~ can be estimated using the *IS*~*p*~/*IS*~*c*~ results. When *K* = 10 and the *IS*~*p*~/*IS*~*c*~ value is around 2.0, this results in *K*~*T*~ = 13, which is in close agreement with the *IS*~*p*~/*IS*~*c*~ value of 2.35 for a band with a total of 13 different band intensities. A detailed explanation of the *K*~*T*~ value can be found in the main text. Although the estimates for *K*~*T*~ tend to be higher than the value that can be estimated with *K*~*T*~ based only on the *IS*~*p*~/*IS*~*c*~ ratio, it is notable that the 2 values are in close agreement for the cases where the *IS*~*p*~/*IS*~*c*~ ratio is two or above. In the second test of the method, we examine the same pattern of noise in a different fashion. The signal of the image in Figure (#F1){ref-type=”fig”}(c) is composed of 2 bands with *I*~*p*~ = 1 and *I*~*c*~ = 2. The noise starts at a