OPTIMIZATION TECHNIQUES OF NUMBER PLATE RECOGNITION ALGORITHM FOR IMPERFECT DETECTION CONDITIONS

  • Hana Stefanović Visoka škola strukovnih studija za IT, Beograd
  • Radosav Veselinović Ekonomski fakultet, Beograd
  • Goran Bjelobaba Narodna banka Srbije, Beograd
  • Ana Savić Visoka škola elektrotehnike i računarstva stukovnih studija, Beograd
Keywords: Automatic Number Plate Recognition (ANPR), Hough transformation, Optical Character Recognition (OCR), imperfect detection

Abstract

This paper provides an optimization of a simple MATLAB-based technique for Automatic Number Plate Recognition (ANPR). Digital image segmentation, after resizing image and removing noise, is applied, while some edge detection algorithms and some morphological techniques are used. Additional spatial filtering and Hough transformation are applied to compensate the variables that can affect the ANPR’s ability to produce an accurate read, such as time of day, weather and angles between the cameras and the license plates. Free Optical Character Recognition (OCR) software is used to output results.
Published
2019-01-15
Issue
Vol 2017 No 64 (2017)
Section
Articles