Number plate recognition using template comparison for various fonts in MATLAB

Abstract

In a diverse country like India, gated communities, corporate premises and university campuses witness a lot of unfamiliar vehicles with number plates in different formats, fonts, font sizes and sometimes even in various languages which enter and leave the estate every day, and it is difficult to register the vehicle numbers manually even for a multi-lingual person. This document aims to extract image frames from a streaming CCTV footage, recognize the vehicle number and convert it into its corresponding text format. Template matching has been used earlier in recognition of digits and letters, this paper also uses the concept of template matching methodology using a SIMULINK model in MATLAB which is developed to extract the vehicle number from the number plate. The alphanumeric values are extracted from the image containing the number plate, one after the other, and are matched with its corresponding template with the help of the template matching methodology. This template matching block compares each and every pixel value of the original image with the pixel values of the corresponding stored template and returns a template metrics value as an output. The templates stored in the model consists or various fonts in English as well as templates of some of the widely used languages (22 official languages) across India. The SIMULINK model uses the Digital Signal Processing Toolbox and the Computer System Vision Toolbox in MATLAB. Various factors affects the efficiency of this process such as luminance, blurred images, incomplete numbers or characters, damaged number plates, etc.