The color of an object is characterized by three reflection densities, measured through three primary color filters. A set of subtractive primaries is, therefore, characterized by a color matrix (c) in which the elements are the densities of unit concentrations of the primaries. The vector density which characterizes a reproduction is obtained from the vector density of the subject by operating on it with the matrix (c). Two successive reproductions are characterized by the matrix product of the two individual matrices. The character of the masking which is required for correct reproduction is determined by the inverse matrix (c-1). A method is described for appraising the results to be achieved by any scheme of partial masking, and these results are applied to successive reproductions such as are involved in reproducing a Kodachrome transparency. The method is applied to calculating the masking which is required in a reproduction to compensate for errors which are anticipated in a subsequent reproduction. The extent to which these matrix calculations are vitiated by the non-additivity of the constituent densities is discussed.
CARL W. MILLER, "The Matrix Algebra and Color Reproduction," J. Opt. Soc. Am. 31, 477-478 (1941)