A bayesian network approach in the relevance feedback of personalized image semantic model

Lei Huang, Jian Guo Nan, Lei Guo, Qin Ying Lin

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

2 Scopus citations

Abstract

Based on a natural and friendly human-computer interaction, relevance feedback is used to determine a user's requirement s and narrow the gap between low-level image features and high-level semantic concepts in order to optimize query result s and perform a personalized search. In this paper, we proposed a novel personalized approach for image semantic retrieval based on PISM (Personalized image semantic model), which use the user queries related to the image of feedback mechanism, dynamic image adjustment semantic similarity of the distribution, and fuzzy clustering analysis, PISM training model to make it more accurate expression of semantic image to meet the different needs of the user's query. And the limitations of image-based semantic memory of learning algorithm, the initial experimental system developed by a number of user feedback to participate in relevant training, which analyzes the performance of the algorithm, the experiments show that the algorithm is a viable theory, with a value of the application.

Original languageEnglish
Title of host publicationAdvances in Multimedia, Software Engineering and Computing Vol.1
Subtitle of host publicationProceedings of the 2011 MSEC International Conference on Multimedia, Software Engineering and Computing, November 26-27, Wuhan, China
EditorsDavid Jin, Sally Lin
Pages7-12
Number of pages6
DOIs
StatePublished - 2011

Publication series

NameAdvances in Intelligent and Soft Computing
Volume128
ISSN (Print)1867-5662

Keywords

  • Bayesian Networks
  • Image retrieval
  • Image semantic
  • Personalized
  • Relevant feedback

Fingerprint

Dive into the research topics of 'A bayesian network approach in the relevance feedback of personalized image semantic model'. Together they form a unique fingerprint.

Cite this