Image Database Management Architecture: Logical Structure and Indexing Methods

摘要

Visual information is an important type of information in modern life. However, it is still not used by organizations in a full capacity. The major reason for that is the lack of internal structure of visual information. The existence of this structure in numerical data allows to build very effective tools for classification, storage, and retrieval of numerical information, such as a relational data management system. In case of visual information, each value of the picture is basically meaningless, but the set of pixels starts carry meaningful information. In this paper, we aim to classify different types of images based on the areas of origination and application. We also suggest the possible structure of the database management system with images as elements of it. Another objective is to propose the indexing methods, which allow to avoid the direct comparison of visual query consequently to entire database. We also introduce the idea of applying multi frame super-resolution method to development of store-retrieval procedures for a database with dynamical visual information.