Details:
Summary:
An earlier work (Lo et. al. 1999a) presented in TAGA'99 was carried out using single stimuli to study color models' performance between a self-luminous reproduced softcopy and the reflection original. The results were used to verify or modify the CIECAM97s color model found performing the best. In practice, however, complex images are frequently used in areas such as graphic arts and desktop publishing. Hence, a set of new experiments is conducted in the viewing conditions of dim surrounds using complex images. Five color models are tested in this study. They were CIELAB, von Kries, LLAB, ZLAB, and CIECAM97s. A flatbed scanner HP 4C/T, characterized under each viewing condition tested by a forward 3rd-SVD (Singular Value Decomposition) masking type of characterization model (Lo et. al. 1999b), was used to obtain image data in terms of RGB format. In addition, the PLCC model was used to characterize the Barco monitor used in this study to display images tested. The work was conducted closely following the CIE guidelines (Alessi 1994). The matching performance of the softcopy processed by a particular color model was judged against a standard (original) illuminated in a viewing cabinet (dim surround) by a panel of observers. Also the results will be used to confirm to the findings obtained using single stimuli and further modify the best performing model found in this study in future works. The final aim of these works alongside a framework successively conducted is to derive a reliable color appearance model, efficiently accounting for disparate media/viewing conditions.