Inverse characterization in a printing device is the process to find control values of colorants to print out any input stimulus values (CIEXYZ, CIELAB, etc.). In a CMY-type printer, the control values can be simply estimated through the interpolation process using a lookup table with a one-to-one relation between the control values and the stimulus values. In a multicolorant printer with extra colorants like red (orange), blue, and green, however, since it has one to many correspondences between CIELAB values and control values, an appropriate control value must be selected from many candidate control values which have negligible color differences from input stimulus value. Selecting a control value without any restriction tends to induce interpolation errors because it does not consider the relation between neighbor control values. In this article, we propose an improved inverse characterization method for multicolorant printer to reduce interpolation errors using the correlation between distributions of control values. We first sampled the CIELAB values regularly in CIELAB space in order to find the appropriate control values for each CIELAB value, since a color stimulus can be represented by several control values of colorants in a multicolorant printer. To find control values for the sampled CIELAB values, the colorant space is sampled and the CIELAB values for all combinations of control values were estimated using the Cellular Yule Nielsen Neugebauer spectral model. The control value whose estimated CIELAB value was close to a sampled CIELAB value was extracted as a candidate for the appropriate control value of the sampled CIELAB value. Subsequently, the most appropriate candidate was selected by considering global and local correlation. For this purpose, we selected all control values for the sampled CIELAB values so that all the selected control values had higher similarity than the predefined threshold in the distribution of colorant amount for global selection step. In addition, in the local selection step, regarding the sampled CIELAB values for which we could not select a control value via this global selection step, the control value was reselected by comparing similarities between the neighbor selected control values and candidates in CIELAB space. Then, accurate CIELAB values of the selected control values were measured and stored in the lookup table. To evaluate the proposed inverse characterization method, a CMYKGO printer was utilized. The proposed method effectively reduced the color difference in the interpolation process. Moreover, the gamut was extended partially and the continuous tone could be represented more smoothly than by conventional methods.
In-Su Jang, Chang-Hwan Son, Tae-Yong Park, Yeong-Ho Ha, "Improved Inverse Characterization of Multicolorant Printer Using Colorant Correlation" in Journal of Imaging Science and Technology, 2007, pp 175 - 184, https://doi.org/10.2352/J.ImagingSci.Technol.(2007)51:2(175)