A generalized technique for studying the information properties of channels and recognition systems

The paper proposes a generalized methodology for studying the information properties of channels and recognition systems, applicable to the analysis of complex recognition systems. The application of the technique showed that when combining heterogeneous channels for recognition, it is effective to use not only direct, but also indirect information about the object’s belonging to classes in the alphabet of final (general) solutions. Combining channels with basic and indirect information increased the amount of information received by 10-40 percent.