Learning feature relevance and similarity metrics in image databases

B. Bhanu, Jing Peng, Shan Qing

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

32 Scopus citations

Abstract

Most of the current image retrieval systems use »one-shot» queries to a database to retrieve similar images. Typically a K-NN (nearest neighbor) kind of algorithm is used where the weights of the features that are used to represent images remain fixed (or manually tweaked by the user) in the computation of a given similarity metric. However, neither all of the features are equally important for a given query nor a similarity metric is optimal for all kinds of images in a database. The manual adjustment of these weights and the selection of similarity metric are exhausting. Moreover, they require a very sophisticated user. The authors present a novel image retrieval system that continuously learns the weights of features and selects an appropriate similarity metric based on the user's feedback given as positive or negative image examples. Experimental results are presented that provide the objective evaluation of learning behavior of the system for image retrieval.

Original languageEnglish
Title of host publicationProceedings - IEEE Workshop on Content-Based Access of Image and Video Libraries, CBAIVL 1998
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages14-18
Number of pages5
ISBN (Print)0818685441, 9780818685446
DOIs
StatePublished - 1998
Event1998 IEEE Workshop on Content-Based Access of Image and Video Libraries, CBAIVL 1998 - Santa Barbara, United States
Duration: 21 Jun 1998 → …

Publication series

NameProceedings - IEEE Workshop on Content-Based Access of Image and Video Libraries, CBAIVL 1998

Other

Other1998 IEEE Workshop on Content-Based Access of Image and Video Libraries, CBAIVL 1998
Country/TerritoryUnited States
CitySanta Barbara
Period21/06/98 → …

Fingerprint

Dive into the research topics of 'Learning feature relevance and similarity metrics in image databases'. Together they form a unique fingerprint.

Cite this