Pattern recognition by indexing

We consider a set‐theoretic approach to pattern recognition and introduce the definition of an index for patterns and inverse patterns. Unlike traditional recognition methods, in our approach pattern identification reduces to operations with inverse patterns. We show that these novel notions let us significantly reduce the computational load by replacing pattern comparison operations with a much smaller number of inverse pattern intersection operations.