Comic Image Category Classification Using SIFT Features

The recent development of the information society leads to many multimedia data on the Web. Especially in Japan, there are a lot of comic images on the Web. However, such a phenomenon causes copyright problems. In order to develop useful information society, the need for comic managing system identifying the author of comic images and ignoring the copied images is increasing. Such a system is indispensable for our future information society. In this paper, we propose a system for identifying the image of comics, and evaluate its performance against conventional methods. Furthermore, we examine the performance with subjective evaluation result by specialists of cartoonist.