A New Image Enhancement and Super Resolution technique for license plate recognition

Abdelsalam Hamdi,Yee Kit Chan,Voon Chet Koo

doi:10.1016/j.heliyon.2021.e08341

Abstract

License Plate Recognition (LPR) is an important implemented application of Artificial Intelligence (AI) and deep learning in the past decades. However, due to the low image quality caused by the fast movement of vehicles and low-quality analogue cameras, many plate numbers cannot be recognised accurately by LPR models. To solve this issue, we propose a new deep learning architecture called D_GAN_ESR (Double Generative Adversarial Networks for Image Enhancement and Super Resolution) used for effective image denoising and super-resolution for license plate images. In this paper, we show the limitation of the existing networks for image enhancement and image super-resolution. Furthermore, a feature-based evaluation metric called Peak Signal to Noise Ratio Features (PSNR-F) is used to evaluate and compare performance between different methods. It is shown that the use of PSNR-F has a better performance indicator than the classical PSNR-pixel-to-pixel (PSNR-pixel) evaluation metric. The results show that using D_GAN_ESR to enhance the license plate images increases the LPR accuracy from 30% to 78% when blur images are used and increases the accuracy from 59% to 74.5% when low-quality images are used.

Highlights

Computer vision applications have been widely used in the past few years
This paper proposes a Double Generative Adversarial Network (GAN) for image Enhancement and Super Resolution (D_GAN_ESR) to improve the LR images using a deep learning system
The proposed Single Image Super-Resolution (SISR) results for License Plate (LP) images are compared with the following SISR methods, SRGAN [21], EDSR [20], and ESRGAN [46], using three evaluation metrics

Summary

Introduction

Computer vision applications have been widely used in the past few years. The capabilities of Artificial Neural Network (ANN) and Convolutional Neural Network (CNN) [1] shown in the past decades made it possible to translate pixels values to actions designed by developers. One of the commonly used computer vision applications is the License Plate Recognition (LPR) [2, 3, 4]. LPR is a computer vision application that automatically recognises the license plate characters from an image and converts them to editable text. LPR has a lot of essential applications nowadays. LPR systems could do that in seconds [5]. LPR is used for intelligent parking systems and many other applications. LPR faces many challenges, as the Optical Character Recognition (OCR) accuracy is proportionally related to the quality of the input image.

Methods

Results

Conclusion