Methods and algorithms for solving problems in the automatic recognition of license plates

Abstract

© 2019, Institute of Advanced Scientific Research, Inc.. All rights reserved. The article shows the relevance of developing methods and algorithms for automatic recognition of license plates in the images. These methods and algorithms are used to control traffic safety and to ensure the operation of car parks in the access control systems of vehicles to protected areas. For the design of these systems, recognition, parsing, localization, normalization and segmentation procedures are carried out. During localization, part of the vehicle license plates is detected in the image. Normalization leads to a change in the size and orientation of the resulting image fragment to a form suitable for further processing. Segmentation allows you to highlight individual characters of the license plates. The recognition result is a text string with identified license plates characters. To determine the elements of a string, parsing is used. These tasks are resource-intensive and, as a rule, must be performed in real time with high accuracy. Therefore, the development of effective methods and algorithms that ensure the achievement of the required time indicators and the accuracy of automatic recognition of license plates for use in various information processing and control systems is relevant. The paper discusses the various methods of solving these problems, the possibility of detection and localization of the searched area in the image based on the method of Viola-Jones and normalization using Hough transform is proposed and the analysis of histograms of brightness and character recognition of numbers using the method of support vector machines.