Content-based Image Retrieval (CBIR): Principles, Concepts and Query Approaches

摘要

Nowadays, digital laboratories are widely used as visual information. With a view to increasing growth of this great complex and inefficiency of images text indexing systems, for efficient retrieval of images, content-based image retrieval (CBIR) systems have been formed more than ten years before. Generally, these systems attempt to retrieve similar images with defined property of user or pattern (such as shape design and image example). Their aim is to support mage retrieval based on its content features (such as shape, color and texture) that are usually propounded as feature vectors. One of the major privileges of CBIR, is the possibility of an automatic retrieval procedure instead of the former method based on keywords, which required the former time consuming and hard annotation of database images. CBIR technology is used in different applications such as The CBIR technology has been used in several applications such as fingerprint identification, digital libraries, crime prevention, medicine, historical research, among others. The aim of this article is introduction of the concepts and problems related to creation of CBIR systems and statement of the two query approaches of QBE (Query by Example) and GBDA (Group Biased Discriminant Analysis). In this article it has been tried to show a general view in relation to these systems and their forming structures.