Sampling optimization for printer characterization by greedy search

Printer color characterization, e.g., in the form of an ICC output profile or other proprietary mechanism linking printer RGB/CMYK inputs to resulting colorimetry, is fundamental to a printing system delivering output that is acceptable to its recipients. Due to the inherently nonlinear and complex relationship between a printing system's inputs and the resulting color output, color characterization typically requires a large sample of printer inputs (e.g., RGB/CMYK) and corresponding color measurements of printed output. Simple sampling techniques here lead to inefficiency and a low return for increases in sampling density. While effective solutions have been proposed to this problem very recently, they either do not exploit the full possibilities of the 3-D/4-D space being sampled or they make assumptions about the underlying relationship being sampled . The approach presented here does not make assumptions beyond those inherent in the subsequent tessellation and interpolation applied to the resulting samples. Instead, the tradeoff here is the great computational cost of the initial optimization, which, however, only needs to be performed during the printing system's engineering and is transparent to its end users. Results show a significant reduction in the number of samples needed to match a given level of color accuracy.