Similarity learning via dissimilarity space in CBIR

Nguyen, G.P.; Worring, Marcel; Smeulders, A.W.M.

doi:10.1145/1178677.1178695

Cited by 20 publications

(13 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some of the approaches proposed so far employed the original definition of dissimilarity space, where dissimilarities are computed by taking into account multiple prototypes of relevant images [12,5]. Other authors have proposed to use the "dissimilarity space" technique for combining different feature space representations [2].…”

Section: From Multi-spaces To Dissimilarity Spacesmentioning

confidence: 99%

“…This approach is based on the creation of a new space where patterns are represented in terms of their (dis)similarities to some reference prototypes. Thus the dimension of this space does not depend on the dimensions of the low-level features employed, but it is equal to the number of reference prototypes used to compute the dissimilarities This technique has been used recently to exploit Relevance feedback in content-based image retrieval field [5,12], where relevant images play the role of reference prototypes. In addition, dissimilarity spaces have been also proposed for image retrieval to exploit information from different multi-modal characteristic [2].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Dissimilarity Representation in Multi-feature Spaces for Image Retrieval

Piras

Giacinto

2011

Image Analysis and Processing – ICIAP 2011

View full text Add to dashboard Cite

Abstract. In this paper we propose a novel approach to combine information form multiple high-dimensional feature spaces, which allows reducing the computational time required for image retrieval tasks. Each image is represented in a "(dis)similarity space", where each component is computed in one of the low-level feature spaces as the (dis)similarity of the image from one reference image. This new representation allows the distances between images belonging to the same class being smaller than in the original feature spaces. In addition, it allows computing similarities between images by taking into account multiple characteristics of the images, and thus obtaining more accurate retrieval results. Reported results show that the proposed technique allows attaining good performances not only in terms of precision and recall, but also in terms of the execution time, if compared to techniques that combine retrieval results from different feature spaces.

show abstract

Section: From Multi-spaces To Dissimilarity Spacesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Dissimilarity Representation in Multi-feature Spaces for Image Retrieval

Piras

Giacinto

2011

Image Analysis and Processing – ICIAP 2011

View full text Add to dashboard Cite

show abstract

“…For that purpose, (dis)similaritybased representations have been recently proposed [15], [17], [19], [20]. As pointed out by these authors, similarities are convenient to manipulate multimodal information since they form a homogeneous representation of the content.…”

Section: A Feature-based Representationmentioning

confidence: 99%

Multimodal Preference Aggregation for Multimedia Information Retrieval

Bruno¹,

Marchand-Maillet²

2009

JMM

View full text Add to dashboard Cite

Abstract-Representing and fusing multimedia information is a key issue to discover semantics in multimedia. In this paper we address more specifically the problem of multimedia content retrieval through the joint design of an original multimodal information representation and of a machine learning-based fusion algorithm. We first define a novel preference-based representation particularly adapted to the retrieval problem, and then, we investigate the RankBoost algorithm to combine those preferences to fullfill a user's query. Interestingly, it ends up being a flexible retrieval model that only manipulates ranking information and is blind to the intrinsic properties of the multimodal information input. The approach is tested on annotated images and on the complete TRECVID 2005 corpus and compared with SVM-based fusion strategies. The results show that our approach equals SVM performance but, contrary to SVM, is parameter free and faster.

show abstract

“…For that purpose, (dis)similarity-based representations have been recently proposed [1,2,9,15]. As pointed out by these authors, similarities are convenient to manipulate multimodal information since they form a homogeneous representation of the content.…”

Section: Feature-based Representationmentioning

confidence: 99%

Combining multimodal preferences for multimedia information retrieval

Bruno

Kludas

Marchand-Maillet

2007

Proceedings of the International Workshop on Workshop on Multimedia Information Retrieval

View full text Add to dashboard Cite

Representing and fusing multimedia information is a key issue to discover semantics in multimedia. In this paper we address more specifically the problem of multimedia content retrieval by first defining a novel preference-based representation particularly adapted to the fusion problem, and then, by investigating the RankBoost algorithm to combine those preferences and a learn multimodal retrieval model. The approach has been tested on annotated images and on the complete TRECVID 2005 corpus and compared with SVMbased fusion strategies. The results show that our approach equals SVM performance but, contrary to SVM, is parameter free and faster.

show abstract

Similarity learning via dissimilarity space in CBIR

Cited by 20 publications

References 27 publications

Dissimilarity Representation in Multi-feature Spaces for Image Retrieval

Dissimilarity Representation in Multi-feature Spaces for Image Retrieval

Multimodal Preference Aggregation for Multimedia Information Retrieval

Combining multimodal preferences for multimedia information retrieval

Contact Info

Product

Resources

About