A Long Term Learning Scheme in CBIR Systems by Defining Semantic Templates Using Information of Similarity-Refinement Based Short Term Learning
Subject Areas : electrical and computer engineeringعصمت راشدی 1 , حسین نظامآبادیپور 2
1 -
2 -
Keywords:
Abstract :
In This paper, a new scheme for long term learning in CBIR systems is proposed. In this scheme, semantic templates are extracted from information provided through relevance feedback process for short-term learning which use similarity refinement techniques. This information will be used as semantic templates in future retrieval sessions to improve the precision of the CBIR system. Also, a similarity function is introduced to calculate the similarity between queries and semantic templates. The proposed method is examined on a database with 10000 color images. The experimental results and comparison with ‘iFind’ method, confirm the effectiveness of the proposed method.
