Image Retrieval through Online Multimodal Distance Metric Learning

Author(s): L. Dinesh, S. Sasikala

Abstract: Content-based image retrieval also known as query by image content (QBIC) and content-based visual information retrieval (CBVIR).”Content-based” means that the search analyzes the contents of the image. Distance metric learning (DML) is a main method to perfect similarity search in content-based image retrieval. In previous Distance metric learning methods in general belong to single-modal DML in that they learn a distance metric either on a single type of feature or on a combined feature space by simply concatenating multiple types of diverse features together. In this method some problems are occur they are 1. Some types of features May significantly dominate the others in the DML task, weakening the ability to exploit the potential of all features and 2. The naive concatenation approach may result in a combined high dimensional feature space, making the subsequent DML task computationally intensive. In this paper a novel framework of Online Multi-modal Distance Metric Learning (OMDML), this learns distance metrics from multi-modal data or multiple types of features via online learning scheme. Our propose multi-modal distance metric learning scheme for content-based image retrieval, which consists of two phases, 1.  Learning phase and 2. Retrieval phase. Learning phase work to facilitate the image ranking task in the retrieval phase. Retrieval phase work to produce the list of corresponding top-ranked images to user. Experimental result to evaluate the performance of our algorithms for multi-modal image retrieval, in which results validate the effectiveness of our proposed technique.