Mathematical Model for Constructing a Map of the Location of Areas that Make Up the Background Environment for Optical-Electronic Systems

Abstract

A developed mathematical model is presented for constructing a map of the location of areas that make up the background environment for optical-electronic systems. A feature of the proposed model is that when determining which class a particular pixel belongs to, the class values of neighboring pixels, as well as the spatial position of the pixel, are simultaneously taken into account. Quantitative assessments of the adequacy of this model and existing mathematical image models capable of constructing maps of the location of areas are given. The proposed model was compared with the Markov model, block Markov model, Gibbs model, block Gibbs model and doubly stochastic model with level quantization. Adequacy assessment was carried out using an artificial neural network developed by the authors, which evaluates the similarity of two images using a normalized similarity index ranging from 0.0 to 1.0, and its k-fold cross-validation. The comparison results showed that the developed model, according to the calculated indicator, is at least three times better than the known ones.