A Web Image Retrieval Re-ranking Scheme with Cross-Modal Association Rules
Yong Zhu, Naixue N. Xiong, Jong Hyuk Park, Ruhan He · 2008
This paper proposes a new re-ranking scheme and presents experimental performance results for Web image retrieval with integrated query. In our previous work, cross-modal association rule was designed for associating one keyword with several visual feature clusters in Web image retrieval. Based on the cross-modal association rule, we implement an automatic re-ranking process online to integrate the keyword and visual features for Web image retrieval, and gives experimental test. The experiment is carried out in a Web image retrieval system named VAST (VisuAl & SemanTic image search), and the results show the effectiveness of the re-ranking method scheme.