@phdthesis{Wyble2007_2,
        Abstract = {The main goal of this paper is to consider methods of analyzing the effect of color measurement instrumentation performance on modern color reproduction systems. This will be accomplished by first exploring characterization models for two output devices. Next, a detailed explanation of evaluation methods for color measurement instrumentation is presented. A framework is now in place to explore the areas and amount to which color measurement instrumentation can affect the predictive abilities of the device models.

The device characterization models presented here are for color halftone printers and DLP projectors. The goal of output characterization models is to predict color output for a given set of device dependent input coordinates. For example, the DLP model accepts RGB input values and predicts CIEXYZ values. In this paper, halftone models are generally derived from the Neugebauer model. Many modern models are based on this important and very successful model. The DLP projector model is a new model based upon the modeling of similar displays, such as LCD and CRT monitors. The DLP projector has the additional complication of a white channel. A new transformation is required to determine the white component which is then added to the traditional red, green, and blue components.

Both device models are dependent on color measurement instrumentations in similar ways. Model parameters are derived from measurements needed for model optimization. These parameters will have some variance due to the variance associated with the measurement instrument. Likewise, when evaluating model performance, comparisons are made between measured data and model predicted data. Again, the instrument data has variance, and this variance will affect the conclusions regarding the accuracy of the device model. Both aspects need to be considered to fully understand the effect of color instrumentation on the model output.

Several methods are proposed to evaluate color measurement instrumentation. These are generally derived from ASTM E2214-02. This specification focuses on historic univariate and new multidimensional methods for evaluation. The result of a multidimensional performance analysis is a covariance matrix or, equivalently, an ellipsoid. This ellipsoid describes the boundary of statistically- expected values for a given measurement system. To apply to common industrial applications of these measurements, two aspects of evaluation are considered: the use of a single instrument (instrument repeatability) and the use of multiple instruments (inter-instrument repeatability). The statistics for these two evaluations are similar, but it is very important to understand the magnitude of the variances of these two methods. Usually, modern instruments make very self-consistent measurements (repeatability) but very often multiple instruments are used for characterizations (for example in different facilities). Depending on their needs, instrument users must perform the appropriate analyses to apply the instrument evaluation to their application.

This paper concludes with a set of future experiments and improvements that can be used to extend this work. The most important of these is a description of methods by which the instrument evaluation data can be propagated through the device models. These propagation techniques will create statistical descriptions of the model parameters as well as the model output values. By understanding these statistical distributions, the effectiveness of the output device models be fully understood.},
        Address = {Chiba, Chiba, Japan},
        Author = {David R. Wyble},
        Keywords = {device characterization},
        Month = {},
        School = {Chiba University, Faculty of Physics, },
        Title = {Color Measurement for Device Characterization},
        Type = {Ph.D. Dissertation},
        Url = {},
        Year = {2007}