Hypergraph-based image search reranking with elastic net regularized regression
- PDF / 2,181,052 Bytes
- 24 Pages / 439.642 x 666.49 pts Page_size
- 39 Downloads / 201 Views
Hypergraph-based image search reranking with elastic net regularized regression Noura Bouhlel1
· Ghada Feki1 · Chokri Ben Amar1
Received: 6 September 2019 / Revised: 12 July 2020 / Accepted: 21 July 2020 / © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Image search reranking is emerging as an effective technique to refine the text-based image search results using visual information. In this paper, we introduce a novel hypergraph-based image search reranking method that accounts for both relevance and diversity of search results. Namely, the text-based image search results are taken as vertices in a probabilistic regression-based hypergraph model and reranking is formulated as a hypergraph ranking problem with absorbing nodes. More specifically, to discover related samples and characterize the relationships among them, we bring the Elastic Net regularized regression model into the hypergraph construction. Exceeding the conventional hypergraph construction schemes, our scheme is able to describe the high-order relationships and the local manifold structure among visual samples while ensuring the datum-adaptiveness. Afterward, we apply a hypergraph-based ranking with absorbing nodes to ensure a diversified reranking. That is, during the reranking process, previously-ranked samples are transformed into absorbing nodes at each iteration, thereby redundant ones are prevented from receiving high ranking scores. Extensive experiments on real-world data from Flickr suggest our proposed reranking method achieves promising results compared to existing reranking methods. Keywords Image search · Visual reranking · Hypergraph · Elastic net · Regularized regression
Noura Bouhlel
[email protected] Ghada Feki [email protected] Chokri Ben Amar [email protected] 1
REGIM: Research Groups in Intelligent Machines, University of Sfax, National Engineering School of Sfax (ENIS), BP 1173, Sfax, 3038, Tunisia
Multimedia Tools and Applications
1 Introduction Owing to the explosive growth of social networks and photo-sharing websites, the number of on-line images has been exponentially increased. As a result, the need for an effective image search system is stronger than ever [6, 16, 17, 30, 31]. Traditional image search systems (e.g., Flickr1 ) are commonly based on textual information. Namely, a user keyword is matched with the textual information around an image (including the title, textual descriptions, user-provided tags, etc) for search and indexing [29]. Nevertheless, such systems cannot comprehensively describe the rich content of images since they totally ignore the visual information. Besides, they suffer from the shortcomings that the textual information is noisy, ambiguous and language-dependent. As a consequence, image search results may contain irrelevant and duplicate items which may significantly affects the search performance [17, 24]. In light of this, image search reranking is introduced as one of the promising techniques to enhance the performance of text-based ima
Data Loading...
