Image recognition is an information process implemented by some information converter (intelligent information channel, recognition system) having input and output. The input of the system is fed with information about the characteristics of the objects being presented. The output of the system displays information about which classes (generalized images) the recognized objects are assigned to. When creating and operating an automated system for pattern recognition, a number of problems are solved, while for different authors the formulations of these tasks, and the set itself, do not coincide, since it depends to a certain extent on the specific mathematical model on which this or that recognition system is based. This is the task of formalizing the domain, forming a training sample, learning the recognition system, reducing the dimensionality of space.