Neural Network Based License Plate Recognition System

The first aim of our methods developed in our structure is, to detect license plate zone from image of vehicle which has standard license plate. Second step is to get noiseless character image from license plate zone. Third step is to recognize the characters from image by neural network based system. The target of this type systems are automatically reading the plate of vehicles and deciding to punishing, collecting toll, controlling and managing without human brain at superhighways, bridges and auto parks etc. At this approach, we used Gabor Filtering, Connected Component Labeling and Adaptive Threshold methods to detect license plate zone. We decide and used our special threshold method, noise remover filter (vertical and horizontal filters) and histogram methods at segmenting characters.