Cross-Language Image Retrieval Using Hitting Time Model

As the resources of images grow day by day on the WEB, the demand of image retrieval from users becomes stronger and stronger. They want to share images with people from different countries. Nowadays, most image retrieval systems is based on single language, thus, it is difficult for different language speakers to share images. In this paper, we propose a novel method of using hitting time model to retrieve images at bilingual environment. The result of experiment shows that this method can efficiently improve the semantic correlation between query and retrieved images at bilingual environment, especially when lacking image resource in the query language.