A novel retrieval framework using classification, feature selection and indexing structure

Feng, Y., Urruty, T. and Jose, J. M. (2010) A novel retrieval framework using classification, feature selection and indexing structure. In: 16th International Multimedia Modeling Conference, MMM 2010, Chongqing, China, 6-8 Jan 2010, pp. 731-736. (doi: 10.1007/978-3-642-11301-7_77)

Abstract

In this paper, we propose a framework to consider both the efficiency and effectiveness to achieve the trade-off in performance of Content Based Image Retrieval (CBIR). This framework includes: (i) concept based classification to classify images into different semantic concept groups and narrows down the search domain in retrieval; (ii) Feature selection model to analysis the relationship between queries and concept classes to reduce feature dimension; (iii) Multidimensional vector space indexing structure for real-time access to reduce the retrieval cost. In our experiments, we study the efficiency and the effectiveness of our method using one public collection and compared with one of state of the art methods.

• Item Type: Conference Proceedings
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Jose, Professor Joemon and Urruty, Mr Thierry and Feng, Mr Yue
• Authors: Feng, Y., Urruty, T., and Jose, J. M.
• College/School: College of Science and Engineering > School of Computing Science