ID | 113909 |
Author |
Fujisawa, Akira
Tokushima University
Matsumoto, Kazuyuki
Tokushima University
Tokushima University Educator and Researcher Directory
KAKEN Search Researchers
Yoshida, Minoru
Tokushima University
Tokushima University Educator and Researcher Directory
KAKEN Search Researchers
Kita, Kenji
Tokushima University
Tokushima University Educator and Researcher Directory
KAKEN Search Researchers
|
Keywords | Image Classification
Illustration Image
Color Histogram
Style
IF-hist
Sensibility
|
Content Type |
Journal Article
|
Description | Illustration images used in comics or animation have a style that constitutes an emotional feature. The style is represented by the various elements of the illustration image. There are a few research studies that treat the style of illustration images. In this paper, we tried to model the style by using an image feature, and to classify an illustration image by style. We thought that the color that appeared infrequently would represent the style of the illustration image. Therefore, we proposed the method for creating the color histogram, which emphasizes colors appearing infrequently. We term this \Infrequency histogram; IF-hist." To test the effectiveness of the histogram, we experimented with classifying two styles defined as “For boys" and “For girls" based on the IF-hist. The results of the experiment indicated that, when using the IF-hist, the precision of the classification result of the style “For girls" is 93%. This is 50% higher than when using the existing normal color histogram. Moreover, the precision of the other classification results has improved.
|
Journal Title |
International Journal of Advanced Intelligence
|
ISSN | 18833918
|
Publisher | AIA International Advanced Information Institute
|
Volume | 8
|
Issue | 1
|
Start Page | 72
|
End Page | 83
|
Published Date | 2016-05
|
EDB ID | |
FullText File | |
language |
eng
|
TextVersion |
Publisher
|
departments |
Science and Technology
|