A trainable retrieval system for cartoon character images

摘要: This paper proposes a novel method to retrieve cartoon character images in database or network. In this method, partial features of an image, defined as regions and aspects, are used keys identify images. The similarities between query image the computed by using these features. Based on similarities, same similar identified retrieved from database. Moreover, our adopts training scheme reflect user's subjectivity. emphasizes significant aspects assigning more weight based preferences actions, such selecting desired area image. These processes make retrieval effective accurate. Experimental results verify effectiveness accuracy method.