Nafez O. Neurolike methods of complex texture images processing in computer networks

The object of study - is the process of digital proccesing difficult textur images in the computers systems on the basis of neurolike methods. The subject of research - the errors of the digital difficult textur images processing of objects of different nature. Methods of investigation are based on using of vehicle Q-conversion, parallell-hierarchical conversion, artificial neurons networks, mathematical logic and programing, graphs, mathematical analysis set and probability theory theory. The aim of the dissertation is to increase of exactness and fast-acting of treatment of difficult textur images on the basis of the specialized automated procedures for segmentation and recognition pathologies. The thesis is is dedicated to elaboration of neuronetwork model of computer processing of texture images, the application of such model stimulated the development of new neurolike computer-based procedures enabling to calculate correlative function with higher accuracy, speed and invariance regarding brightness change, contrast, scale, structural-functional organization, as elaboration of recommendations on architecture and structural-functional organization of Pi memory to provide hight efficiency of complex texture images conversion. Practical value of the results obtained is the following creation of experimental base needed for elaboration of methods of compact presentation and recognition of complex textural images. Degree of implemention is in medical sphere and in education process.