A Color Transformation for the Compression of Cmyk

ABSTRACT

A CMYK image is often viewed as a large amount of device-dependent data ready to be printed. In several circun-

stances, CMYK data needs to be compressed, but the conversion to and from device-independent spaces is imprecise

at best. In this paper, with the goal of compressing CMYK images, color space transformations were studied. In

order to have a practical importance we developed a a new transformation to a YYCC color space, which is device-

independent and image-independent, i.e. a simple linear transformation between device-dependent color spaces. The

transformation from CMYK to YYCC was studied extensively in image compression. For that a distortion measure

that would account for both device-dependence and spatial visual sensitivity has been developed. It is shown that

transformation to YYCC consistently outperforms the transformation to other device-dependent 4D color spaces

such as YCbCrK, while being competitive with the image-dependent KLT-based approach. Other interesting con-

clusions were also drawn from the experiments, among them the fact that color transformations are not always

advantageous over independent compression of CMYK color planes and the fact that chrominance subsampling is

rarely advantageous.

1. INTRODUCTION

Images for monitor display are commonly stored or transmitted in popular color spaces such as RGB, CIELAB,

YUV, YCbCr, etc..1,2 Images to be printed are commonly rendered in a subtractive color space such as CMYK.1,2

Sometimes documents contain pictures in other color spaces which are then converted to CMYK for printing.

It is often necessary to compress CMYK images for a number of reasons, such as bandwidth or memory reduction.

Fig. 1 shows a printing system path in which a CMYK image bitmap is generated and possibly compressed. JPEG3

is a standard for image compression and is widely used. Other compressors such as SPIHT4 are becoming very

popular and work is underway to define a new standard for still-image compression.5 In this paper we use only

JPEG for three basic reasons. First, JPEG is a well-known standard so we assume the readers are well aware

of its performance. Second, our tests involve comparisons whose results are not likely to change by replacing the

coding mechanism. Last, as a standard, it is the coder in which most readers would actually use the results of this

paper. We here study color transformation for compression of CMYK. Similar to the way in which RGB images are

commonly converted to luminance-chrominance color spaces such as YCbCr before compression, we would like to

understand which transformations would be the best to apply to CMYK images for the same purpose. Along with

color transformations we also discuss the applicability of subsampling for chrominance planes.

Most of the contents of this paper are also available in its Journal version.6

2. DEVICE-DEPENDENCE FOR COLOR SPACES

Color values and color spaces are dependent on a particular imaging or input device for which they were prepared or

from which they were acquired. Good color management practice demands color correction for the printing device,7,8

invlving both gamut mapping procedures as well as color correction. We will refer to both steps as simply color

correction. When preparing CMYK data from image data in a known color space to image data to be imaged in

a particular printer, the color correction process not only maps the input gamut to the device's gamut but it also

incorporates other factors such as under-color removal (UCR) and tone-reproduction curves (TRC). Those steps are

commonly based on a printer model7,8 which is empirically derived. CMYK is generally a device-dependent color

space. CMYK here refers to device-CMYK where the device is known. The