Online Non-feedback Image Re-ranking via Dominant Data Selection
Chen Cao, Shifeng Chen, Yuhong Li, Jianzhuang Liu
Image re-ranking aims at improving the precision of keywordbased image retrieval, mainly by introducing visual features to re-rank. Many existing approaches require offline training for every keyword, which are unsuitable for online image search. Other real-time approaches demand user interaction, which are inappropriate for large-scale image collection. To improve the accuracy of web image retrieval in a practicable manner, we propose a novel re-ranking algorithm to explore the cluster information of the image set. First, we build spectral graph on images that retrieved by search engine, and remove isolated nodes as noisy images. Then, we select positive samples from the most dominant cluster in initial top-ranked images, and the samples are used for semi-supervised learning and ranking. Our algorithm is online and non-feedback. Experiments on two public databases demonstrate that our algorithm outperforms the state-of-the-art approaches.
Figure 1: The framework of our method.
Figure 2: Four categories in INRIA database. Top-15 ranked images using (a) the search engine and (b) our approach. Non-class images are in red box.
Figure 3: INRIA database: query mean precision over 353 categories. (From left to right: Search Engine, LabelDiag, SpecFilter and Ours).
Figure 4: INRIA database: MAP of re-ranking results over 353 categories. (From left to right: Search Engine, LogClass (textual), LogClass (visual), LogClass (hybrid t+v), MRank, LabelDiag, SpecFilter and Ours).
ü  C. Cao, S. Chen, Y. Li, and J. Liu, “Online Non-feedback Image Re-ranking via Dominant Data Selection,” Proc. ACM Multimedia (ACM-MM), 2012. [pdf]