STRUCTURAL SYNTHESIS OF THE COMPUTER VISION AND ITS PARAMETRIC IDENTIFICATION WITH STATISTICAL ESTIMATION OF VARIATIONAL PARAMETERS

Abstract

The article presents the results of a study of the problem on structural synthesis of a computer vision and its parametric identification. We present the main classes of procedural transformations of images and describe a method for constructing a model of a vision system for measuring a deterministic feature using procedural and parametric transformations of modified descriptive image algebras. The function of measuring a deterministic quantitative characteristic of an object of observation is defined as a superposition of a procedural and parametric descriptive algebraic image transformation scheme. The parametric identification problem is formulated in the form of a nonlinear optimization problem. We give examples of the use of the described method in measuring the area of objects of various nature. Recommendations are formulated for the statistical processing of the values of the measured characteristic in order to clarify its value.