NEW THEORY OF PATTERN RECOGNITION ON THE BASIS OF STOCHASTIC GEOMETRY

Nikolay Fedotov, Penza State University, fedotov@diamond.stup.ac.ru
Luydmila Shulga, Penza State University, ec@diamond.stup.ac.ru

ABSTRACT

The article offers a new approach towards the construction of recognition features independent of images’ displacement or linear deformation. The distinguishing characteristics of the group of features under study is representing each of them as a sequential composition of three functionals acting upon the function of one variable. The process to construct the new features suggested boasts of the advantages as follows: a)a host of new features can be easily constructed; b) the features obtained can be structurized along with parallel computations. Great many new features have been constructed to successfully solve the task of recognizing coloured images in biological systems, for instance, blood cells in gematology. Keywords: pattern recognition, stochastic geometry, triple functional.