Photo-inkjet Printing Method Based on the Limited Total Amount and Dot-visibility of Six-colorant

With the rapid growth of photo-inkjet printers using sixcolorant in response to the spread of digital and mobile cameras, there has been a lot of research regarding six-color separation in order to achieve visually smooth tones in highlight regions. Most conventional method use the maximum amount of colorant with a low dot-visibility, thus resulting in paper wetness, poor edgesharpness, and ink spreading and blotting in printed images. In order to reduce the amount of unnecessary colorants, this paper proposes a photo-inkjet printing method based on the limited total amount and dot-visibility ordering of a six-colorant. First, CIELAB values are estimated for an input CMY image using a color-mixing model, then these values are compared with the precalculated CIELAB values that correspond to all combinations of the CMYKlclm (Cyan, Magenta, Yellow, blacK, light cyan, and light magenta) colorants. This is done by using a color-difference constraint, which can determine the initial CMYKlclm candidates. Next, limitations on the total amount of colorant are imposed on the initial CMYKlclm candidates in order to remove excessive colorant, then the final CMYKlclm candidates are determined by minimizing the usage of light cyan and light magenta in the dark regions, based on a dot-visibility ordering of C, M, Y, K, lc, and lm. Finally, a lookup table is generated. It is composed of the CMY values versus the CMYlclm digital values. Experiments showed that the proposed method could effectively reduce the use of unnecessary colorants, while preserving good image quality. Introduction With portable cameras and LCD displays becoming increasingly popular, interest in photo-quality printing has also rapidly increased, thus resulting in the development of hi-fi printers with more than four colorants, such as CMYKOG (Cyan, Magenta, Yellow, blacK, Orange, and Green) and CMYKlclm [1]. CMYKOG printers are generally used in the textile industry to extend color gamut and reduce metamerism by adding extra colorants, i.e. orange and green with different hues from C, M, Y, and K [2]. Meanwhile, CMYKlclm printers focus on reproducing smooth tones in highlight regions by using light cyan and light magenta, which have the same hues as that of standard colorants, yet different concentrations. Moreover, the use of various concentrations can reduce coarse dot patterns and achieve greater resolution on various substrates, such as uncoated, copy, and photo papers. The CMYKlclm printers have gradually become a favorite of customers in the current market with regard to making inroad into film developing and printing. In high-fidelity printer with the N-primary colorants, the (N-3) degrees of freedom, equal to the number of extra colorants, makes it possible for reproducing the same tristimulus value by different combination of colorants amount, with changing the gamut shape and dot-visibility as an essential part of image quality. The degree of freedom raises the redundancy problem and thus a six-color separation method is first required to compute the amount of CMYKlclm colorants that correspond to the CMYK colorants when CMYKlclm colorants are printed on papers using digital halftones. Traditional six-color separation using color-difference identifies monotonically increasing functions that convert M into (M, lm) and C into (C, lc), under the constraints of minimizing color-difference [3]. Although this method yields accurate colorimetric reproductions, the image quality is inferior due to the use of cyan or magenta that has high dot-visibility in bright regions. Hauang and Nystrom presented a method that produced a smooth tone and decreased abrupt differences in lightness between light and dark colors, color-differences, however, arose due to hue differences between diluted and saturated colorants [4]. As such, the six-color separation method using additional colorants and a quantitative granularity metric was also proposed by Son et al [5], where yellow and light magenta with a low dot-visibility were used as additional colorants based on a color-mixing rule in bright regions. As a result, differences in hue were corrected and the amount of graininess was reduced. Meanwhile, Agar proposed a model-based color separation method in order to reduce the amount of measurement data and to maximize the usage of colorants via an ordering of Y, lc, lm, C, M, and K [6]. Here, an accurate colorimetric reproduction and a smooth-dot pattern can be simultaneously achieved, yet excessive amounts of colorants are used due to the maximum usage of colorants with low dotvisibility. Such factors can result in paper wetness, an increase in dry time, and degradation of image quality due to poor edgesharpness, ink spreading and blotting. Accordingly, in order to reduce the amount of unnecessary colorants, this paper proposes a photo-inkjet printing method based on the limited amount and dotvisibility ordering of a six-colorant. For an objective assessment, the CIE1976 color-difference metric, the total amount of color, and dot-visibility were evaluated. Regarding a subjective evaluation using real images, the resulting images were compared with other method. Six-colorant Separation Based on Total-ink Limitation and Dot-visibility Ordering A flowchart of the proposed method for reducing the amount of colorant used is shown in Fig. 1. First, CMY and CMYlclm digital values, that satisfy uniform data distribution in CIELAB space, are generated and printed using digital halftones. Their spectral reflectance is then estimated using the YNSN (Yule Nielsen Spectral Neugebauer) model in order to reduce the measurement data [7]. Next, the estimated spectral reflectance is multiplied with the relative daylight power distribution and a color-matching function to create CIELAB values. These values are compared by using the constraint of a color-difference less 15th Color Imaging Conference Final Program and Proceedings 255 3 than that of a pre-defined threshold. This results in initial CMYlclm candidates corresponding to each CMY digital value. Thereafter, excessive amount of colorants are removed by imposing a limitation on the total amount of colorant for the initial CMYlclm candidates, then the final CMYlclm candidates are obtained by reducing the usage of lc and lm in dark regions, based on a dot-visibility ordering of C, M, Y, K, lc, and lm. Finally, a lookup table, composed of the CMY versus CMYlclm digital values, is constructed to simplify the complex computations of the six-color separation method. Generation of digital values for CMY and CMYlclm Printer characterization based on YNSN Extraction of initial candidates using color difference metric Extraction of final candidates for the reduction of excessive amount of colorants Construction of the CMY:CMYlclm lookup table Figure 1. The flowchart of the proposed six-color separation method Generation of Digital Values for CMY and CMYlclm The CMY and CMYlclm digital values, stored in the lookup table for six-color separation, are generated by uniform and nonuniform sampling. Regarding calibrated CMY digital values using a linearization method, the combination of CMY digital values sampled based on an interval of 51 = n for each channel creates a uniform CIELAB value distribution, however, if the digital values for the CMYlclm colorants are generated using uniform sampling, the CIELAB values will congregate in dark regions, while the number of sample points in the bright regions is seriously deficient. Therefore, the digital values for the CMYlclm colorants are non-uniformly sampled based on an interval of 11 = n on average in bright regions with fine sampling and a 128 ~ 0 range, and 21 = n in dark regions with a 255 ~ 129 range. Estimation of Spectral Reflectance Using Color Mixing Model Depending on the printer, the CMY and CMYlclm digital values need to be converted into an independent color space for accurate color reproduction. Since increased amounts of colorants are used in hi-fi printers, an established model-based color mixing model, the YNSN model, is used to estimate the spectral reflectance of arbitrary printed patches based on the weighted average spectral reflectance of the Neugebauer primaries [7]. At the same time, it is important to consider the optical scattering into the substrate, which can be achieved by the following: