Artificial intelligence techniques in retrieval of visual data semantic information

The development of the information society results in the fact that an ever increasing amount of information is stored in computer databases, and a growing number of practical activities depend on efficient retrieval and association of the data. In the case of textual information, the problem of retrieving the information on a specific subject is comparatively simple (although it has not been fully solved from a scientific point of view). On the other hand, the application of databases that store multimedia information, particularly images, causes many more difficulties. In such cases, the connection between the subject-matter of the content (i.e. the meaning of the image) and its form is often very unclear; while the retrieving activities as a rule aim at the image content, the accessible methods of searching refer to its form. With the aim to partially solve those emerging problems this paper presents new opportunities for applying linguistic algorithms of artificial intelligence to undertake tasks referred to by the authors as the automatic understanding of images. A successful obtaining of the crucial semantic content of an image thanks to the application of the methods presented in this paper may contribute considerably to the creation of intelligent systems that function also on the basis of multimedia data. In the future the technique of automatic understanding of images may become one of the effective tools for storing visual data in scattered multimedia databases and knowledge based systems. The application of the automatic understanding of images will enable the creation of automatic image semantic analysis systems which make it possible to build intelligent multimedia data retrieval or interpretation systems. This article proves that structural techniques of artificial intelligence may be useful when solving a given problem. They may be applied in the case of tasks related to automatic classification and machine perception of semantic pattern content in order to determine the semantic meaning of the patterns. This article paper presents ways of applying such techniques in the creation of web based systems and systems for retrieving and interpreting selected medical images. The proposed approach will be described in selected examples of medical images obtained in radiological and MRI diagnosis, however the methodology under consideration has general applications.