An Interval Approach to Pattern Recognition of Numerical Matrices
Alexander Prolubnikov · 2013
We present an interval approach for pattern recognition of numerical matrices. We construct systems of interval linear equations that are as-sociated with given numerical pattern matrices. Considering a system of interval linear equations as a family of systems of real linear equations, we use a measure of variation of these systems solutions as a measure of distance between matrices. As an application, we use our technique for the recognition of raster images that are distorted in the course of noising. The results of computational experiments are presented.