In order to find desired pictures by color keywords on the Internet, it is necessary to extract color information from a picture database considering perceptual color. However, there are different viewing conditions such as monitor difference, different viewing distance and so on. This means that a color extraction robot or a retrieval engine must have the allowance of color ambiguity, considering perceptual color under different viewing conditions when they are realized. For the purpose of clearing the relationship between perceptual colors under different viewing conditions and controlled conditions, we modify the collaborative picture evaluation system. Experiments was opened to the students in our university through the Internet, and continued for 2 weeks to get 16 voluntary testers. According to experimental result using 50 fruit or vegetable pictures and 40 color representations, it is found that three objective colors (red, orange, brown) disagree with the perceptual colors.
[1]
Takayuki Hamamoto,et al.
The Collaborative Evaluation Method of Coded Picture Quality Using the Internet
,
2000
.
[2]
C.-C. Jay Kuo,et al.
A semantic classification and composite indexing approach to robust image retrieval
,
1999,
Proceedings 1999 International Conference on Image Processing (Cat. 99CH36348).
[3]
Seiichiro Hangai,et al.
Collaborative method for quality evaluation of coded pictures
,
1999,
Proceedings 1999 International Conference on Image Processing (Cat. 99CH36348).
[4]
Andrea Kutics,et al.
An object-based image retrieval system using an inhomogeneous diffusion model
,
1999,
Proceedings 1999 International Conference on Image Processing (Cat. 99CH36348).