¹Computer Engineering Department, Zhongshan Polytechnic, Zhongshan, China
²Shenzhen Angelshine Co., LTD, Shenzhen, China

---

Abstract—Color space transformation is the process of converting color from one space to the other type of color space, which plays a very important role in image acquisition, display and color information transfer. Though CMYK space is applied to express color when using printer or phototypesetter, editing correction in CMYK color space may cause color loss and lower the speed of computer operation. Thus, it is often converted into CIE L*a*b* uniform color space which is device-independent. In general, there existed a problem of that the dark tone has a big color difference in previous methods when doing color transformation. In order to handle such issue, we put forward a sampling method for color space with the nonuniform segmentation, which utilizes multiple regressions, prism interpolation and pyramid interpolation to model the color space for the same color inkjet. The technical difficulties of the three methods and their implementations are introduced in detail, together with the realizations from CMYK space to CIE L*a*b* space. The comparisons among such three methods are carried out by horizontalvertical way. By analyzing the modelling results, it clearly shows that the proposed way for color space transformation can get better effect by using polynomial regression method and three-dimensional modelling method. The experiment results also indicate that the introduced approach is a kind of good converting way for color space with less color difference, better effect and much simpler, and it can be easily made widespread into other color space transformation.

Index Terms—color space transformation, multiple regression, prism interpolation, pyramid interpolation, color difference

[PDF]

Cite: Jiabin Deng, Juanli Hu, and Juebo Wu, "A Study of Color Space Transformation Method Using Nonuniform Segmentation of Color Space Source," Journal of Computers vol. 6, no. 2, pp. 288-296, 2011.