Uncovering the Effect of Visual Saliency on Image Retrieval

Visual saliency modeling has achieved impressive performance for boosting vision-related systems. Intuitively, it should be beneficial to content-based image retrieval task, since the users' query attention is heavily related to the region of interests (ROI) in query image. Although some approaches have been proposed to combine image retrieval systems with visual saliency models, no a comprehensive and systematic study is made to discover the effect of different saliency models on image retrieval in a qualitative and quantitative manner. In this paper, we attempt to concretely investigate the diversity of visual saliency models on image retrieval by making extensive experiments based on nine popular saliency models. To cooperatively mining the complementary information from different models, we also propose a novel approach to effectively involve visual saliency into image retrieval systems by a learning process. Extensive experiments on a generally used image benchmark demonstrate that the new image retrieval system remarkably outperforms the original one and other traditional ones.