Correction of License Plates Recognition Data by Templates Matching

Automatic license plate recognition (ALPR) is generally considered as a solved problem, but despite significant progress developed algorithms are essentially designed for recognition in constrained conditions. This study provides an algorithm for templates matching based recognition of located and segmented into characters license plate (LP) image in one stage. That is not by splitting it into two separate stages of defining template type in the first step followed by subsequent recognition of LP. The algorithm first classifies segmented characters and then corrects obtained data through defining their editing operations with minimal total cost for the purpose of matching in result with one of template types or reference text lines in database. This research is towards developing actual ALPR of multinational LPs carried out.