A self-relevance feedback method based on object labels

User's relevence feedback is often included in many content-based image retrieval (CBIR) systems, and this method is proved to be effective in improving the retrieval result. However, it may cause too much user participation which may make users impatient. To solve this problem, the paper propo...

Full description

Saved in:
Bibliographic Details
Main Authors: Jiabin Ruan, Yubin Yang
Format: Conference Proceeding
Language:English
Subjects:
Boats
concept occurrence vector
Image Retrieval
self-relevance feedback
semantic
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:User's relevence feedback is often included in many content-based image retrieval (CBIR) systems, and this method is proved to be effective in improving the retrieval result. However, it may cause too much user participation which may make users impatient. To solve this problem, the paper proposes a self-relevance feedback method for CBIR which needs no user involvement. Self-relevance is seldom mentioned in CBIR as it is usually difficult to increase the performance of a system. Based on the "concept occurrence vector" (COV) used for image retrieval, the proposed method can improve the precision of the retrieval process, which is proved by our experiments. Though the improvement is not very huge, the method make the application of self-relevance feedback in CBIR possible.
ISSN:2161-9069
DOI:10.1109/ICCASM.2010.5620002