Derivation of Digital Matrices without Conversion Errors

Abstract

Conversion between the signal sets of GBR and YC may often introduces errors in the converted signals, when digital matrices are used. This is due to the finite bit-length of the matrix coefficients and the rounding of the converted signal to a finite bit-length. The paper presents a derivation method in which teh error-free digital matrices are obtained, as well as an example of such matrices for the CCIR Rec. 709 colorimetry.