Visual experiments on four displays (two LCD, one CRT and hardcopy) were conducted to determine colorimetric tolerances of images systematically altered via three different transfer curves. The curves used were: Sigmoidal compression in L*, linear reduction in C*, and additive rotations in hat>More than 30 observers judged the detectability of these alterations on three pictorial images for each display. Standard probit analysis was then used to determine the detection thresholds for the alterations. It was found that the detection thresholds on LCD's were similar or lower than for the CRT's in this type of experiment. Summarizing pixel-by-pixel image differences using the 90th percentile color difference in AE*ab was shown to be more consistent than similar measures in AE94 and a prototype AE2000. It was also shown that using the 90th percentile difference was more consistent than the average pixel wise difference. Furthermore, SCIELAB pre-filtering was shown to have little to no effect on the results of this experiment since only global color-changes were applied and no spatial alterations were used. Acknowledgements The author wishes to thank all those who helped make this thesis a reality: My many patient observers without whom, none of this would be possible. Dr. Mark Fairchild for his expertise and encouragement. Mitch Rosen for his advice, support and encouragement, even when things looked dim. Sharron, Sergio, Arturo, Lawrence, Scott and Sun Ju-my fellow students for the many interesting questions, discussions and otherwise friendly banter during the past two years. My family for putting up with me for another two years. . . sorry I had to leave so soon. My extraordinary wife Sarah who's love and encouragement have brightened my life more than she'll ever know. A Special thanks to the IBM T.J. Watson Research Center for their kind support and funding of this research project. Come to the edge. "It's too high" Come to the edge! So I did And you pushed And I Flew. . .
[1]
Mark D. Fairchild,et al.
The ZLAB Color Appearance Model for Practical Image Reproduction Applications
,
1998
.
[2]
Roy S. Berns,et al.
Performance testing of color‐difference metrics using a color tolerance dataset
,
1989
.
[3]
M. Luo,et al.
Chromaticity‐discrimination ellipses for surface colours
,
1986
.
[4]
Mark D. Fairchild,et al.
COLOR-APPEARANCE REPRODUCTION: VISUAL DATA AND PREDICTIVE MODELING
,
1999
.
[5]
Karen M. Braun,et al.
Viewing techniques for cross‐media image comparisons
,
1996
.
[6]
H. Ifill,et al.
High-Information-Content Color 16 . 3 ”-Desktop-AMLCD with 15 . 7 Million a-Si : H TFTs
,
1998
.
[7]
Mark D. Fairchild,et al.
Colorimetric characterization of the Apple studio display (Flat panel LCD)
,
1998
.
[8]
Manuel Rubiño,et al.
Considerations on the calibration of color displays assuming constant channel chromaticity
,
1995
.
[9]
K. Witt,et al.
Parametric variations in a threshold color‐difference ellipsoid for green painted samples
,
1983
.
[10]
Susan Farnand.
The Effect of Image Content on Color Difference Perceptibility
,
1996,
Color Imaging Conference.
[11]
Mark D. Fairchild,et al.
Considering the surround in device-independent color imaging
,
1995
.