The normalization and regularization of terminology being used in the subject domain "Mathematical Theory of Pattern Recognition and Image Analysis" is a crucial and critical problem for theory and applications of modern Pattern Recognition and Image Analysis. There are, at least, two fields where the absence of widely accepted and logically ordered terminology is a serious obstacle to further development and activity: automation of pattern recognition and image analysis and conference paper reviewing.

       For the previous we need a standard regular language providing an opportunity to present uniformly pattern recognition/image analysis transforms and data/image models. The descriptive image algebras provide such a language. The second precondition is availability of a logically ordered classifications pattern recognition/image analysis algorithms and data/image descriptors. To construct necessary classifications we need a thesaurus of the subject domain and eventually an appropriate ontology.

       Badly-organized lists of thematic topics in international conferences on pattern recognition and image analysis and low quality of paper reviews, caused by the problem of choosing appropriate domain experts, also indicate that the given subject domain requires a regular linguistic mean for supporting these activities. It is well known, that a thesaurus-based ontology is a useful mechanism for solving these problems.

       Ontologies have been proposed as an important means for information processing, analysis and integration. They are widely used in many areas, such as knowledge management and organization, natural language processing, databases and knowledge bases, digital libraries, geographic information systems, information retrieval (including, content-based image retrieval), automation of problem solving in the domain of image processing, analysis and recognition and so on. Ontology is a system that describes concepts and relations between them in some domain of interest.

       In order to structure domain knowledge and to unify terminology it is necessary to develop an ontology of the subject domain "Mathematical Theory of Pattern Recognition and Image Analysis". The ontology will allow efficiently describe semantics of domain data, solve the problem of contradictory terms and limit possible interpretations of terms.

       At present time, there are no thesauri and ontologies of the given domain. Since the overwhelming majority of the literature on pattern recognition and image processing, analysis and understanding is written in English and there are also a lot of publications on mathematical theory of pattern recognition and image analysis in Russian, it is appropriate to construct a thesaurus with terms and definitions both in English and in Russian. The thesaurus will be a base for the ontology of the given domain.

       The ontology development process will include the following steps.

Step 1. Investigation into subject domain "Mathematical Theory of Pattern Recognition and Image Analysis"

Content. Selection and expert estimation of information sources on pattern recognition and image processing, analysis and understanding. Investigation into knowledge system and logic of the subject domain. Identification and definition of key concepts and relationships in the domain of interest and the terms that refer to such concepts.

Duration.18 months.

Results. Vocabulary of terms related to pattern recognition and image processing, analysis and understanding.

Step 2. Development of the thesaurus on pattern recognition and image analysis.

Content. Development of the thesaurus structure. Construction of classification schemes for the terms. Specification of concepts meaning (creation of reconcilable definitions for each basic concept). Construction of term records. Final verifying and testing of the thesaurus by experts in the given domain.

Duration. 18 months.

Results. Thesaurus on pattern recognition and image analysis.

Task 3. Development of the domain ontology on the basis of the thesaurus.