Review and research on "semantic gap" problem in the content based image retrieval
Guohua Geng · Journal of Northwest University · 2005
Aim To discuss the gap problem which exists in the CBIR.Methods The form and origin of the problem are explored;from the point of acquiring the image semantics,the methods and their current lacks on solving the problem are studied and analyzed,and towards them some resolving strategies are presented elementarily.Results At present the applying multilayer relevance feedback methods in the CBIR could build and modify the association low-level feature and high-level semantic,which would be helpful to narrow the gap in image retrieval and reach the purpose of semantic retrieval at some extent.Conclusion To achieve the real semantic-based image retrieval would be the most effective approach to settle the problem.