| dc.description.abstract | The color mismatch between display devices is a relevant issue in color-critical
applications. Here we use a colorimetric display calibration technique based on
visual judgments that do not need a measurement device. This process is divided
into two main stages, the data collection by psychophysical experiments and the
computation of a matrix-based transformation. The transformation is applied to the
color values of a display so it reproduces the desired color appearance. We designed
a GUI in JavaScript to collect a number of hue judgments that will describe the
display color characteristics. Before the hue selection task, we determined the
neutral gray of the display, which is set as a background for the experiments. From
the data is created a calibration file which saves the color coordinates of the observer
selections, the hue planes, the neutral grey, and the signal-to-light-intensity function.
From this file, we computed three types of matrix-based transformations that were
tested for two pairs of theoretical displays and two sets of hue combinations. The
results show variations in the transformation performance; however, we could point
out some results. The proposed transformation to preserve the greyscale achieves
its purpose, being more noticeable when there is a larger color difference between
displays. From the color difference after each transformation appears two groups
of observers. The observers working in the color science field show a closer color
matching after the transformations. Last, from the analysis of two hue sets (four
unique hues; and three unique hues plus a binary hue), we obtained similar results
for both hue combinations. We argue that it is feasible to add binary hues to the
calibration method. If similar results are obtained for binary and unique hues,
it would be desirable to choose between hues with less inter-observer variability,
which produces better color mapping, and make the task easier, for example, by
being faster to choose. | |