Robust Multi-view Manifold Ranking for Image Retrieval
- 2.4k Downloads
Graph-based similarity ranking plays a key role in improving image retrieval performance. Its current trend is to fuse the ranking results from multiple feature sets, including textual feature, visual feature and query log feature, to elevate the retrieval effectiveness. The primary challenge is how to effectively exploit the complementary properties of different features. Another tough issue is the highly noisy features contributed by users, such as textual tags and query logs, which makes the exploration of such complementary properties difficult. This paper proposes a Multi-view Manifold Ranking (M2R) framework, in which multiple graphs built on different features are integrated to simultaneously encode the similarity ranking. To deal with the high noise issue inherent in the user-contributed features, a data cleaning solution based on visual-neighbor voting is embedded into M2R, thus called Robust M2R (RM2R). Experimental results show that the proposed method significantly outperforms the existing approaches, especially when the user-contributed features are highly noisy.
KeywordsImage retrieval Multi-view learning Manifold ranking Data cleaning
The authors would like to thank the anonymous reviewers for their constructive suggestions. This work was supported in part by the ‘Natural Science Foundation of China’ (61301185, 61370070 and 61300071), the ‘Fundamental Research Funds for the Central Universities’ (2015JBM029), and the ‘Science Foundation of Beijing Jiaotong University’ (2015RC008).
- 1.Blum, A., Mitchell, T.: Combining labeled and unlabeled data with co-training. In: COLT, pp. 92–100 (1998)Google Scholar
- 3.He, J., Li, M., Zhang, H., Tong, H., Zhang, C.S.: Manifold-ranking based image retrieval. In: ACM Multimedia, pp. 9–16 (2004)Google Scholar
- 4.Hoi, S., Liu, W., Chang, S.F.: Semi-supervised distance metric learning for collaborative image retrieval. In: CVPR, pp. 1–7 (2008)Google Scholar
- 5.Jain, V., Varma, M.: Learning to re-rank: query-dependent image re-ranking using click data. In: Proceedings of the 20th International Conference on World Wide Web, WWW 2011, pp. 277–286 (2011)Google Scholar
- 7.Pan, Y., Yao, T., Mei, T., Li, H., Ngo, C.W., Rui, Y.: Click-through-based cross-view learning for image search. In: SIGIR, pp. 717–726 (2014)Google Scholar
- 14.Wu, J., Lin, Z., Lu, M.Y.: Asymmetric semi-supervised boosting for SVM active learning in CBIR. In: ACM CIVR, pp. 182–188 (2010)Google Scholar
- 17.Wu, J., Lu, M.Y., Wang, C.L.: Collaborative learning between visual content and hidden semantic for image retrieval. In: ICDM, pp. 1133–1138 (2010)Google Scholar
- 19.Xu, B., Bu, J., Chen, C., Wang, C., Cai, D., He, X.: EMR: a efficient manifold ranking model for content-based image retrieval. IEEE Trans. Knowl. Data Eng. 27(1), 102–114 (2014)Google Scholar
- 24.Zelnik-Manor, L., Perona, P.: Self-tuning spectral clustering. In: NIPS, pp. 1601–1608 (2004)Google Scholar
- 27.Zhou, D., Weston, J., Gretton, A., Bousquet, O., Scholkopf, B.: Ranking on data manifolds. In: NIPS, pp. 169–176 (2003)Google Scholar
- 30.Zhou, Z.H., Zhan, D.C., Yang, Q.: Semi-supervised learning with very few labeled training examples. In: AAAI, pp. 675–680 (2007)Google Scholar