ABSTRACT

Multimedia technologies have been developing rapidly over the last few years and have yielded a large number of databases containing graphical documents. Tools for content-based search of graphical objects have been the subject of intensive research, but their performance is still unsatisfactory for many applications, opening up afield for further research and technology development. Up till now, all popular Internet search engines have been only text-based, including those that search for images. In this paper We propose an image retrieval system based on neural networks. The advantage of using the neural network is that the amount of semantic gap can be reduced when compared to other techniques which may be based on clustering. The methodology proposed below is designed for a specific class of objects, which can be broken down into sub-objects in such a way that the main object can be classified by shape, color distribution and texture of the sub objects and the spatial spatial relations between the sub-objects in a 2-dimensional image. We also assume that translation, scaling and 2D rotation do not change the class of the object, but we do not consider 3Dtransformation.Therefore, photos of the same 3D object from different positions for example are considered to be objects belonging to different class.