Linear Unmixing of colour pigment data

Hi all,

I am a student studying colour pigment fingerprints and how we can improve the identification of such pigments through image reflectance. More specifically, the estimation of each pure pigment in a mixture. I am analysing this data in the range of 400nm-1000nm.

From what I can understand in literature, in the VIS-region, the reflectance data would represent the pigmentation of the minerals or components on the surface, whereas in the NIR-region we would start to see representations of the vibrating molecules.

My question is the following, when two pure elements are mixed together homogeneously with a 50:50 mass ratio, does the linearity hold throughout the observed range? (400-1000nm) Or is the linearity stronger at the reflectance/absorbance bands of both pure elements?

Any insight is appreciated especially that which can be sourced.

Thanks :slight_smile: