Data projectors are often used in demanding imaging applications requiring accurate color. To properly control the color output of such a device, one needs accurate color control models. This paper will describe a color management algorithm for a four-color DLP projector. Four-channel color displays have only recently been introduced to the market. The displays being examined in this paper have the traditional red, green and blue channels and also a supplemental white channel. In parallel to the four-color printer problem, a fourth channel in a display creates a color reproduction challenge since one XYZ can potentially be mapped to many RGBW combinations. A further complication of these projectors is that at the computer interface they are treated as RGB displays. The conversion from RGB to RGBW takes place internally making them at once compatible with current RGB display signals and yet unfriendly to simple color management approaches. The characterization of a display forms the foundation of a mapping from device digital coordinates to colorimetry. This is referred to as the “ forward model.” A common method for characterizing typical RGB color displays starts with three one-dimensional input look-up tables (LUTs) for linearizing the digital input signals with respect to tristimulus values (XYZ). This is followed by a 3x3 matrix for scalar rotation, completing the transformation to tristimulus space. Extending this model to a four-channel display is straightforward: the fourth channel needs its own linearization LUT and the rotation matrix becomes a 3x4. If the RGBW channels of these projectors were all directly controllable, then this would represent the forward model. Unfortunately, the computer can only present RGB digital coordinates that are internally converted to RGBW. Thus, the forward model must include the conversion from RGB to RGBW. To complete the color management of a projector, an inverse model is needed to convert from colorimetry back to device digital coordinates. For these displays, such a model will need to solve both the one-to-many problem of XYZ to RGBW as well as the transformation from RGBW back to RGB. For the balance of this paper, a description of the data collection and a review of the forward model are followed by derivation and use of the inverse model. There are several published efforts regarding the use and characterization of projection displays. Some of these papers have described characterizations of LCD-based projectors. Stone has performed studies on the details of implementing multiple-projector systems. Much theoretical research has been done on the design and modeling of DLP systems. A forward characterization model was demonstrated. The current paper, describing the derivation and implementation of the inverse model, completes the picture of DLP color management.
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
Twelfth Color Imaging Conference: Color Science and Engineering Systems, Technologies, Applications (2004) 228-232
RIT – Main Campus