Improving the accuracy of estimating the measure of proximity between objects during recognition

Abstract

The article analyzes the problem of the occurrence of errors in measuring the geometric parameters of objects in pattern recognition, possible ways to minimize them, and as a result of this analysis, an algorithm is proposed for effectively increasing the reliability of pattern recognition by adequately estimating the measure of proximity between recognized and standard objects. The article considers a model for the process of estimating the measure of proximity between objects and the resulting errors, proposes conditions and an algorithm for minimizing these errors, and recommends measures to improve the accuracy of estimating the measure of proximity between objects.