An image retrieval system with automatic query modification

Aggarwal, Geetika; Ashwin, T. V.; Ghosal, S.

doi:10.1109/tmm.2002.1017734

Cited by 67 publications

(41 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…using the classic Ullman algorithm [13] or more advanced ones. There are algorithms that are automatic or semiautomatic with ability to present preliminary results to the user who may choose important regions and repeat the query [14]. Another group of CBIR algorithms allows queries without full knowledge about searched images or objects.…”

Section: Image Retrieval Algorithmsmentioning

confidence: 99%

The Scalable Distributed Two-layer Content Based Image Retrieval Data Store

Deniziak¹,

Michno²,

Krechowicz³

2015

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

Abstract-The multimedia databases are becoming more and more popular nowadays. One of their main problem is a huge data amount storage. Another problem with multimedia databases is querying. Traditional approaches, based on textual keywords are not sufficient. More advanced techniques, incorporating image content features, should be used. In this paper we propose new multimedia database structure with ability of Content Based Image Retrieval which is based on our previous work: Query by Shape method (QS). Query by Shape is a method which is based on decomposing an object into features. Each feature may consists of shape primitive, a color or a texture. In this paper we only use shape primitives. In order to achieve high scalability and workload control, we propose a modified Scalable Distributed Two-layer Data Structure, as a storage. The modification incorporates adding tree structure, comparing algorithm and returning a set of results to the client.

show abstract

Section: Image Retrieval Algorithmsmentioning

confidence: 99%

The Scalable Distributed Two-layer Content Based Image Retrieval Data Store

Deniziak¹,

Michno²,

Krechowicz³

2015

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

show abstract

“…Later works on relevance feedback and feature re-weighting [1,2,11,24] also assume that features can be represented in fixed-length vectors. Following this assumption the components of the features are all placed into a single vector.…”

Section: Related Workmentioning

confidence: 99%

“…However, these approaches lack the ability to extract as much information from the user feedback as systems with more elaborate user interfaces. Some systems [1,3,21] incorporate multi-class feedback to obtain more information via user feedback. The drawback in using multi-class feedback is the burden on the user having to judge the degree of relevance or non-relevance of each returned image.…”

Section: Related Workmentioning

confidence: 99%

“…This set of images should contain more relevant images than the previous result set. Existing approaches for feature re-weighting via relevance feedback assume that the feature data can be represented in fixed-length vectors, where the corresponding elements of the vectors of each image are located in the same vector positions [1,9,11,12,15,18,21]. Following this assumption, feature re-weighting can be seen to have the effect illustrated in Fig.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Confidence interval approach to feature re-weighting

Lee

Chung

Lee

et al. 2008

Multimed Tools Appl

View full text Add to dashboard Cite

Relevance feedback is commonly incorporated into content-based image retrieval systems with the objective of improving retrieval accuracy via user feedback. One effective method for improving retrieval performance is to perform feature re-weighting based on the obtained feedback. Previous approaches to feature re-weighting via relevance feedback assume the feature data for images can be represented in fixed-length vectors. However, many approaches are invalidated with the recent development of features that cannot be represented in fixed-length vectors. In addition, previous approaches use only the information from the set of images returned in the latest query result for feature reweighting. In this paper, we propose a feature re-weighting approach that places no restriction on the representation of feature data and utilizes the aggregate set of images returned over the iterations of retrieval to obtain feature re-weighting information. The approach analyzes the feature distances calculated between the query image and the resulting set of images to approximate the feature distances for the entire set of images in the database. Two-sided confidence intervals are used with the distances to obtain the information for feature re-weighting. There is no restriction on how the distances are calculated for each feature. This provides freedom for how the feature representations are structured. The experimental results show the effectiveness of the proposed approach and in comparisons with other work, it is shown that our approach outperforms previous work.

show abstract

“…Both query reweighing and query shifting apply a nearest-neighbor sampling approach to refine query concept. Specifically, query reweighing [1][2][3] assigns a new weight to each feature of the query, and query shifting [4][5][6] moves the query to a new point in the feature space. Query expansion [7,8] uses a multipleinstance sampling approach to select samples from the neighborhood of the positive labeled instances for learning.…”

Section: Introductionmentioning

confidence: 99%

A Radial Basis Function and Semantic Learning Space Based Composite Learning Approach to Image Retrieval

Shkurko

2007

2007 IEEE International Conference on Acoustics, Speech and Signal Processing - ICASSP '07

View full text Add to dashboard Cite

This paper introduces a composite learning approach for image retrieval with relevance feedback. The proposed system combines the radial basis function (RBF) based lowlevel learning and the semantic learning space (SLS) based high-level learning to retrieve the desired images with fewer than 3 feedback steps. User's relevance feedback is utilized for updating both low-level and high-level features of the query image. Specifically, the RBF-based learning captures the non-linear relationship between the low-level features and the semantic meaning of an image. The SLS-based learning stores semantic features of each database image using randomly chosen semantic basis images. The similarity score is computed as the weighted combination of normalized similarity scores yielded from both RBF and SLS learning. Extensive experiments evaluate the performance of the proposed approach and demonstrate our system achieves higher retrieval accuracy than peer systems.Index Terms-Radial basis function, semantic learning space, content-based image retrieval

show abstract

An image retrieval system with automatic query modification

Cited by 67 publications

References 25 publications

The Scalable Distributed Two-layer Content Based Image Retrieval Data Store

The Scalable Distributed Two-layer Content Based Image Retrieval Data Store

Confidence interval approach to feature re-weighting

A Radial Basis Function and Semantic Learning Space Based Composite Learning Approach to Image Retrieval

Contact Info

Product

Resources

About