Data correlation obtained by two transformation.

Novakovic, Marijana ; Mandic, Lidija ; Kurecic, Maja Strgar 等

1. INTRODUCTION

Digital image processing involves different devices. Color output devices can be classified into two many types: additive and subtractive (Berns, 2000). Additive color systems produce color on a dark background through the combination of differently colored lights, known as primaries. The additive primaries are red, green and blue (RGB). Displays are the example of additive system (Schein, 1993). Color in subtractive systems is produced through a process of removing (subtracting) in wanted spectral components from white. Typical subtractive are based on cyan, magenta and yellow (CMY). It is common to use a fourth black (K) colorant that allow producing of dark colors. There are RGB color spaces for scanners, cameras, displays and other input devices, and there are CMYK color spaces for printers. In the applications of colorimetry, it is often necessary to transform the values in one set of primaries to those in another set of primaries (Sharma, 2003). Color reproduction usually requires transformation between RGB to CMYK. Adobe Photoshop CS3 is well known as one of the best graphic applications used for image processing at the moment. It works with multiple color models (Rodney, 2005). Together with RGB color space it also supports CMYK. Colorimetric color reproduction include devices that have different principle of color mixing (Morivic, 2008). This means that for a given device signal, the colorimetric coordinates of the image element produced are known with a reasonable degree of accuracy and precision. With colorimetric color reproduction, a user can put together a system in which an image is scanned, the data are converted to colorimetric coordinates (e.g., CIE XYZ), and then these coordinates are transformed into appropriate RGB signals to display on LCD, or into CMYK signals for output to a printer.

The aim of research was to identify the differences among transformations in Photoshop and MatLab. MatLab is interactive mathematical application which is used for mathematical calculations because of the growing importance of color science in manufacturing industry (Westland & Ripamonti, 2004).

2. EXPERIMENTAL PART

The test target, consists of 52 color fields, was made in Adobe Photoshop CS3. Test target was generated in RGB color space. The basic 13 colors were then changed by the characteristics of a hue, lightness and saturation. Every characteristic was not changed more than the 15 units of the basic color. The second column on a test target has only the characteristic of the hue changed, the third column has only the characteristic of the lightness changed and the fourth column has only the characteristic of the saturation changed. The test target was then transformed in CMYK color space in Photoshop. The CMYK values in MatLab are obtained as it shown bellow:

 function [C,M,Y,K] = RGBtoCMYK(R,G,B)

 C = 1 - ( R / 255 )
 M = 1 - ( G / 255 )
 Y = 1 - ( B / 255 )

 K = 1

 if (C < K) K = C
 elseif ( M < K ) K = M
 elseif ( Y < K ) K = Y

 else
 C = ( C - K ) /( 1 - K );
 M = ( M - K ) /( 1 - K );
 Y = ( Y - K ) /( 1 - K );

 C = C*100;
 M = M*100;
 Y = Y*100;
 K = K*100;

 end

The CMYK values were then compared.

3. DISCUSSION

The results are shown in the figures below. On every graph the differences in coordinates between cyan, magenta, yellow and black are shown. The differences are obtained with subtraction of the values gained in MatLab from the values gained by Photoshop. In Tab. 1 are given color patches (letter D is for dark, and L for light color).

In fig. 1. basic, unmodified color fields are shown. The biggest difference can be seen with content of the black color. The difference is negative which shows that the content of the black color in Photoshop is lower, especially for the colors HKS35E, HKS59E, HKS67E, HKS41E and HKS39E. The content of yellow color for colors HKS6E and HKS67E is higher for Photoshop calculations.

In fig. 2. the fields that are modified over hue characteristic are shown.. The biggest differences appear also with the content of black color. Content of magenta is bigger for B color (blue) for Photoshop calculations. The content of yellow color is bigger for colors HKS35, HKS13E, HKS32E and HKS29E.

In fig. 3. the fields that are modified over lightness characteristic are shown. The biggest differences appear also for content of black color, especially for colors G (green), B (blue), HKS6E, HKS35E, HKS59E, HKS67E, HKS41E and HKS39E for MatLab calculations. The yellow color content is bigger for colors HKS35E, HKS32E, HKS59E and HKS29E for MatLab calculations but for the colors HKS6E and HKS67E the content for yellow color is bigger for Photoshop.

In fig. 4. the fields that are modified over saturation characteristic are shown. The content of black colour is the biggest for HKS6E, HKS35E, HKS59E, HKS67E, HKS41E, HKS29E and HKS39E. The yellow color content is also bigger for MatLab calculations except for colors HKS6E and HKS67E. The colors with bigger yellow content in MatLab calculations are HKS35E, HKS32E, HKS41E and HKS29E.

The results shows that is higher portion of black in red and blue color spectra for Photoshop calculations, and magenta in blue part of spectra for MatLab calculations. The figure 4 shows the color fields with modified lightness characteristic. In the blue and green part of spectra fot MatLab calculations the content of the black color is the biggest.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

4. CONCLUSION

The aim of this research was to compare data obtained by two transformations. In Photoshop, calculations are more complex than calculation we made in MatLab. The results show that there are no big differences among these data. The differences were mainly in content of black and yellow portion in given colors. The difference in CMYK values depends on color. The yellow color content is bigger in violet-blue part of the spectra for MatLab calculations as well, except for the light-green part of spectra where calculations in Photoshop are bigger. Some aberration could be noticed with colors that varied in lightness. Our future work will be oriented to find correlation between these CMYK values and values on print.

5. REFERENCES

Berns R.S. (2000), Billmeyer and Saltzman's Principles of Color Technology, John Wiley & Sons, New York, 2000

Morovic, J. (2008). Color Gamut Mappings, John Willey&Sons, ISBN 978-0-470-03032-5, Spain

Rodney, A. (2005). Color Management for Photographers, Focal Press, ISBN 0-240-80649-2, UK

Schein, L. B. & Beardsley, G. (1993). Journal of Imaging Science and Technology, 5.

Sharma, G. (2003). Digital Color Imaging Handbook, CRC Press LLC, ISBN 0-8493-0900-X, USA

Westland, S. & Ripamonti, C. (2005). Computational Color Science using MatLab, John Wiley&Sons, ISBN 0-47084562-7, 2004, UK

Tab. 1. Color patches

 Color Color

R red HKS 4E yellow
G green HKS 59E cyan
B blue HKS 67E Green-L
HKS 6E Green-D HKS 41E gray
HKS 35E violet HKS 29E magenta
HKS 13E orange HKS 39E blue
HKS 32E rose