Efficient Indexing, Color Descriptors and Browsing in Image Databases

This work provides an experimental evaluation of various existing approaches for some of the major problems content based image retrieval applications are faced with. More specifically, global color representation, indexing and navigation methods are analyzed and insight is provided regarding their efficiency and applicability. Furthermore this paper proposes and evaluates the combined use of FastMap and kd-trees to enable accurate and fast retrieval in image databases