LEARNING IN CONTENT BASED IMAGE RETRIEVAL : A REVIEW

Renuka Devi S M

Abstract: Relevance feedback in Content Based Image Retrieval is an interactive process where the user provides feedback on the systemretrieved images to bridge the gap between user semantics at high level and machine extracted low level features of images. RF exploits Machine Learning and Pattern Recognition techniques for Short Term Learning and Long Term Learning to provide improved performance in retrieval. Intra query and across query learning have received enormous attention over the past decade. This paper first categorizes the various learning techniques and discusses the intuition behind each of these techniques. State-of-art learning techniques ranging from Feature Relevance learning to manifold learning in STL and Latent Semantic Analysis used in text processing to recent kernel semantic learning in LTL are discussed

Keywords: Relevance Feedback, Short Term Learning, Long Term Learning, Sematic Gap, High Level Features.

DOI: https://doi.org/10.15623/ijret.2016.0502018