Leveraging non-relevant images to enhance image retrieval performance

Published 1 December 2002

T. V. Ashwin, Rahul Gupta, Sugata Ghosal

Citations5

TL;DR

A robust technique is proposed that utilizes non-relevant images to efficiently discover the relevant search region and a similarity metric, estimated using the relevant images is then used to rank and retrieve database images in the relevant region.

Abstract

Inherent subjectivity in user's perception of an image has motivated the use of relevance feedback (RF) in the image desigined output's retrieval process. RF techniques interactively determine the user's query concept, given the user's relevance judgments on a set of images. In this paper we propose a robust technique that utilizes non-relevant images to efficiently discover the relevant search region. A similarity metric, estimated using the relevant images is then used to rank and retrieve database images in the relevant region. The partitioning of the feature space is achieved by using a piecewise linear decision surface that separates the relevant and non-relevant images. Each of the hyperplanes constituting the decision surface is normal to the minimum distance vector from a non-relevant point to the convex hull of relevant points. Experimental results demonstrate significant improvement in retrieval performance for the small feedback size scenario over two well established RF algorithms.

Keywords

Computer Science

Support vector machine active learning for image retrieval

1,333 Citations2001Simon Tong, Edward Yi Chang

This work proposes the use of a support vector machine active learning algorithm for conducting effective relevance feedback for image retrieval and achieves significantly higher search accuracy than traditional query refinement schemes after just three to four rounds of relevance feedback.

Content-based image retrieval with relevance feedback in MARS

730 Citations2002Yong Rui, Thomas S. Huang +1 more

Experimental results show that the image retrieval precision increases considerably by using the proposed integration approach, and the relevance feedback technique from the IR domain is used in content-based image retrieval to demonstrate the effectiveness of this conversion.

Research Showcase @ Carnegie Mellon University (Carnegie Mellon University)MindReader: Querying Databases Through Multiple Examples

447 Citations2018Yoshiharu Ishikawa, R. Subramanya +1 more

The goal is to provide a user-friendly, but theoretically solid method, to handle ‘diagonal’ queries, and to propose a novel method to “guess” which attributes are important, which correlations areImportant, and with what weight.

View PDF

Small sample learning during multimedia retrieval using BiasMap

296 Citations2005Xiang Sean Zhou, Thomas S. Huang

This paper proposes biased discriminant analysis and transforms specifically designed to address the asymmetry between the positive and negative examples, and to trade off generalization for robustness under a small training sample.

Comparing discriminating transformations and SVM for learning during multimedia retrieval

101 Citations2001Xiang Sean Zhou, Thomas S. Huang

It is argued that relevance feedback problem is best represented as a biased classification problem, or a (1+x-class classification problem), and Biased Discriminant Transform (BDT) is shown to outperform all the others.

Exploring the nature and variants of relevance feedback

41 Citations2002Xiang Sean Zhou, Thomas S. Huang

The nature of the relevance feedback problem in a continuous representation space in the context of multimedia information retrieval is analyzed and the assumptions, implementations, and merits of various solutions in the literature are compared.

Elsevier eBooksAdaptable Similarity Search using Non-Relevant Information

11 Citations2002Ashwin T.V., Rahul Gupta +1 more

A robust technique is proposed to incorporate non-relevant information to efficiently discover the feasible search region and experiments on simulated and benchmark datasets demonstrate robustness and superior performance of the proposed technique over existing adaptive similarity search techniques.

Leveraging non-relevant images to enhance image retrieval performance

5 Citations2002T. V. Ashwin, Rahul Gupta +1 more