Details:
Summary:
In the application of color management system, color transformations between color imaging devices through the medium of CIE colorimetry, practically, use look-up tables, as they will normally be more computationally efficient than a physical mathematical basis. The aim of this study was, hence, to develop an image-processing algorithm efficiently to perform color transformations for printing devices via look-up tables (LUTs). Three processes of color transformation were implemented and evaluated. Those are: 1) scanner to printer, 2) monitor to printer, and 3) printer to monitor. Works subsequently accomplished include to:
- Derived a well-performing printing characterization model carried out both forward (from colorant amounts) and also reverse transforms (into colorant amounts). This is the 2nd-order-SVD (singular-value decomposition) BPA applying polynomial regression, in terms of test target of IT8.7/3 produced using a printing device in interest;
- Generated look-up tables (LUTs) using various different data-size approaches through refined models above;
- Evaluated prediction performances of different LUT data-size approaches for each of 3 imaging devices used (monitor, scanner, and mainly printer);
- Performed actual color transforms of complex images, using derived LUTs which gave reasonably prediction results, with tetrahedral interpolation. Additionally, a weighted-average spatial filtering method was implemented with convolution masks for noise removal.
- Conducted advanced evaluations of models' performances using those complex images produced across media, via forced-choice paired-comparison psychological experiments.
Performances of characterization models derived were investigated using the corresponding training data set (in terms of DE of CIE94 or CIE2000). Also, the GCR (Gray component replacement) approach was carried out in reverse transforms of printing model to both reduce the use of three pigments in producing gray components and enhance the detail of shadow areas of reproductions.