Skip to Main Content
After the research of the principle of SVM, it is applied to the relevance feedback of the content-based image retrieval. First do a initial retrieval separately by MEPG-7 Dominating Color, Gray-level Co-occurrence Matrix and histogram, then do a feedback using the SVM for the new round retrieval. In the experiment we comparing the SVM feedback result with the result of weight adjusting-based algorithm, prove that SVM feedback can greatly improve the recall level, get more items that be of interest to the user.
Date of Conference: 17-19 Nov. 2010