Automated License Plate Recognition: A Survey on Methods and Techniques

Jithmi Shashirangana,Heshan Padmasiri,Charith Perera,Dulani Meedeniya

doi:10.1109/access.2020.3047929

Jithmi Shashirangana, Heshan Padmasiri + Show 2 more

Open Access

https://doi.org/10.1109/access.2020.3047929

Copy DOI

Journal: IEEE Access	Publication Date: Dec 29, 2020
Citations: 155	License type: CC BY 4.0

Affiliation: University of Moratuwa, Cardiff University

Abstract

With the explosive growth in the number of vehicles in use, automated license plate recognition (ALPR) systems are required for a wide range of tasks such as law enforcement, surveillance, and toll booth operations. The operational specifications of these systems are diverse due to the differences in the intended application. For instance, they may need to run on handheld devices or cloud servers, or operate in low light and adverse weather conditions. In order to meet these requirements, a variety of techniques have been developed for license plate recognition. Even though there has been a notable improvement in the current ALPR methods, there is a requirement to be filled in ALPR techniques for a complex environment. Thus, many approaches are sensitive to the changes in illumination and operate mostly in daylight. This study explores the methods and techniques used in ALPR in recent literature. We present a critical and constructive analysis of related studies in the field of ALPR and identify the open challenge faced by researchers and developers. Further, we provide future research directions and recommendations to optimize the current solutions to work under extreme conditions.

Highlights

Automatic License Plate Recognition (ALPR) systems are attracting an increasing interest due to their applicability in intelligent transportation systems that have been installed in many countries for tasks such as traffic law enforcement and traffic monitoring
Unless for ALPR systems, this task requires a sizable amount of labour, time, and resources
We have explored the above 100 related research articles and online resources covering the entire subject area, which were published over the last two decades

Summary

INTRODUCTION

Automatic License Plate Recognition (ALPR) systems are attracting an increasing interest due to their applicability in intelligent transportation systems that have been installed in many countries for tasks such as traffic law enforcement and traffic monitoring. Kim et al [49] have proposed a new algorithm based on colour texture for object detection and demonstrated with a license plate localization system They have extended the previous studies [50], [51] on texture classification by following a Support Vector Machine (SVM) based approach for identifying plate regions. DEEP-LEARNING TECHNIQUES According to the recent development in computer vision approaches, most of the statistical methods have been replaced by deep learning neural networks due to their high accuracy in object detection Embracing this fact, many studies in license plate detection have used different types of FIGURE 4. In [52], Selmi et al, have presented a localization method using a Convolutional Neural Network (CNN) In their study, they have followed two major steps in the license plate detection stage. The model performs well in poor illumination conditions and occlusions

LICENSE PLATE RECOGNITION

PERFORMANCE EVALUATION APPROACHES

Findings

DISCUSSION

CONCLUSION