CONTENT BASED IMAGE RETRIEVAL USING CLASSIFIER ENSEMBLE APPROACH

摘要

Present invention provides specially a study and application of content based image retrieval using classifier ensemble approach. The purpose of the above system is to fuse multiple individual CBIR system to have an accurate and robust CBIR. The proposed method creates a pool of individual system by extracting various distinct features which captures color, texture and geometry features and by applying different feature matching techniques. The propose work tries to apply application orientation fusion of systems, since for different application different specific features would be fused. The selection of systems from the pool of individual systems would be performed based on probabilistic framework of fusion. The different application on which the proposed work is tested upon are OCR (optical character recognition), logo identification, face detection. The proposed work also tries different strategies to speed-up the search space by applying unsupervised data clustering and by utilizing the relevance feedback. The relevance feedback helps in fast convergence of the query search. Following invention is described in detail with the help of Figure 1 of sheet 1 showing the block diagram of proposed work.