Conference Paper

Dynamic Two-Stage Image Retrieval from Large Multimodal Databases

Department of Electrical and Computer Engineering, Democritus University of Thrace, University Campus, 67100 Xanthi, Greece
DOI: 10.1007/978-3-642-20161-5_33 Conference: Advances in Information Retrieval - 33rd European Conference on IR Research, ECIR 2011, Dublin, Ireland, April 18-21, 2011. Proceedings
Source: DBLP


Content-based image retrieval (CBIR) with global features is notoriously noisy, especially for image queries with low percentages of relevant images in a collection. Moreover, CBIR typically ranks the whole collection, which is inefficient for large databases. We experiment with a method for image retrieval from multimodal databases, which improves both the effectiveness and efficiency of traditional CBIR by exploring secondary modalities. We perform retrieval in a two-stage fashion: first rank by a secondary modality, and then perform CBIR only on the top-K items. Thus, effectiveness is improved by performing CBIR on a ‘better’ subset. Using a relatively ‘cheap’ first stage, efficiency is also improved via the fewer CBIR operations performed. Our main novelty is that K is dynamic, i.e. estimated per query to optimize a predefined effectiveness measure. We show that such dynamic two-stage setups can be significantly more effective and robust than similar setups with static thresholds previously proposed.

Download full-text


Available from: Konstantinos Zagoris
  • Source
    • "They represent images with multiple points in a feature space in contrast to single point global feature representations. While local approaches provide more robust information, they are more expensive computationally due to the high dimensionality of their feature spaces and usually need nearest neighbors approximation to perform points matching[18] [19]. Several important features that can be used in IR will be elucidated in the next subsections. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Image retrieval from databases or from the Internet needs an efficient and effective technique due to the explosive growth of digital images. Image retrieval is considered as an area of extensive research, especially in content based image retrieval (CBIR). CBIR retrieves similar images from large image database based on image features, which has been a very active research area recently. The content, that can be derived from image such as color, texture, shape…etc., are called features. This paper will present a survey and discuss the current literature of different types of image retrieval (IR) systems. An overview of the important techniques in image retrieval will be discussed. Finally, some urgent challenges in IR, that have been raised recently, will be presented as well as possible directions for future research.
    Full-text · Article · Jan 2015
  • Source
    • "By this definition, anything ranging from an image similarity function to a robust image annotation engine falls under the purview of CBIR. But, the 'weak spot' of CBIR is that it seems to be notoriously noisy for image queries of low generality [2]. If the query image happens to have a low generality, early rank positions may be dominated by spurious results. "
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper tackles the problem of the user's incapability to describe exactly the image that he seeks by introducing an innovative image search engine called TsoKaDo. Until now the traditional web image search was based only on the comparison between metadata of the webpage and the user's textual description. In the method proposed, images from various search engines are classified based on visual content and new tags are proposed to the user. Recursively, the results get closer to the user's desire. The aim of this paper is to present a new way of searching, especially in case with less query generality, giving greater weight in visual content rather than in metadata.
    Full-text · Conference Paper · Jan 2012
  • Source
    • "We evaluated on the top-1000 results with MAP, precision at 10 and 20. We tested the results for statistical significance against the text-only baseline; image retrieval based on the text queries and annotations was found to perform much better, with a wide margin, than CBIR-only in the same setup [2]. For measuring efficiency, we report the average matching time per topic. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The Bag-Of-Visual-Words (BOVW) paradigm is fast becoming a popular image representation for Content-Based Image Retrieval (CBIR), mainly because of its better retrieval effectiveness over global feature representations on collections with images being near-duplicate to queries. In this experimental study we demonstrate that this advantage of BOVW is diminished when visual diversity is enhanced by using a secondary modality, such as text, to pre-filter images. The TOP-SURF descriptor is evaluated against Compact Composite Descriptors on a two-stage image retrieval setup, which first uses a text modality to rank the collection and then perform CBIR only on the top-K items.
    Full-text · Conference Paper · Jan 2011
Show more