Robust Spectral Implementation of High-Fidelity Printer Characterization for Cross-Media High-Dynamic-Range Imaging Applications

Abstract

This research introduces a multi-spectral type of High-Fidelity (Hi-Fi) colorant color separation algorithm for printing/printer system using more than four inks (colorants). This Hi-Fi colorant separation algorithm, derived for six-colorant CMYKOG printer systems, finds the best colorant combination to reproduce spectra of every color, concerned in high-dynamicrange of original images for an accurate color-matching or – reproduction application. This work revised and extended a previous research, which carried out a seven-ink of CMYKRGB FM printing characterization of presses, only used in the conventional Graphic Arts. An Epson 9900 inkjet printer, having the capability of a six-colorant set, CMYKOG, was spectrally characterized for High-Fidelity (Hi-Fi) multi-color printer simulation. The superset of CMYKOG was divided into three subsets, including two five-colorant groupings of CMYKO and CMYKG, and one four-colorant grouping of CMYK. The spectral printer characterization applied a polynomial regression approach with singular value decomposition method (SVD). The mean ΔE*00 values were 0.98∼1.17 and 1.06∼1.33 found in the forward and the reverse processes respectively. Also 0.0008 ∼ 0.0011 and 0.0009 ∼ 0.0024 of the mean RMSE values were obtained for the forward and the reverse spectral transform respectively. It implies the spectral reflectance of every color in test datasets was satisfactorily reconstructed. Moreover, to robustly perform a feasible approach of spctral color management for Hi-Fi printer simulation in cross-media color reproduction system, a GUI-based of HDR (High-Dynamic-Range) image-fusion module were also be introduced to produce original HDR complex color images. It integrated algorithms of: 1) a multi-raw-processed-image fusion process based on both luminance and gray-balancing strategies; 2) a multiple-conversing-points type of gamut-mapping; 3) Adaptive Multi-illuminants white-balancing; 4) the CIECAM02 color appearance modeling; 5) a Gaussian pyramid type of image enhancement; and 6) a logarithmic HDR tone mapping; and finally an ideal simulated CIEXYZ type of spectra sensing model which adopts interim AdobeRGB connection reference space, combined with CIE 1931 2° standard observer. Consequently, a robust spectral type of cross-media color reproduction system, based on a colorimetric approach of HDR and Hi-Fi (High-Fidelity) was implemented in the study.