Chupikov A. Models of multithreshold image segmentation

The thesis is devoted to development of models and methods providing on the basis of double-step procedures synthesis of partitions of a field of view which are adequate from applied standpoint. The first stage consists in production of partial image segmentation on application of multithreshold processing. The second one is connected with obtaining on the transformation basis of partitions or coverings of image carriers families, which are suitable for problem-oriented interpretation, and correlate with significant objects of the scene. The implemented theoretic and experimental researches of multithreshold models and operations with classes of equivalences or tolerances, in view of their spatial properties and aprioristic feature information connected with specificity of subject-oriented fields, have allowed to create the software providing as solutions of specific processing and interpretation of the visual information problems and their use at a choice of multithreshold segmentation algorithms parameters.