A Deep Learning Framework of Super Resolution for License Plate Recognition in Surveillance System

Recognizing low-resolution license plates from real-world scenes remains a challenging task. While deep learning-based super-resolution methods have been widely applied, most existing datasets rely on artificially degraded images, and common quality metrics poorly correlate with OCR accuracy. We con...

Full description

Saved in:
Bibliographic Details
Main Authors: Pei-Fen Tsai, Jia-Yin Shiu, Shyan-Ming Yuan
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Mathematics
Subjects:
Online Access:https://www.mdpi.com/2227-7390/13/10/1673
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850257672910143488
author Pei-Fen Tsai
Jia-Yin Shiu
Shyan-Ming Yuan
author_facet Pei-Fen Tsai
Jia-Yin Shiu
Shyan-Ming Yuan
author_sort Pei-Fen Tsai
collection DOAJ
description Recognizing low-resolution license plates from real-world scenes remains a challenging task. While deep learning-based super-resolution methods have been widely applied, most existing datasets rely on artificially degraded images, and common quality metrics poorly correlate with OCR accuracy. We construct a new paired low- and high-resolution license plate dataset from dashcam videos and propose a specialized super-resolution framework for license plate recognition. Only low-resolution images with OCR accuracy ≥5 are used to ensure sufficient feature information for effective perceptual learning. We analyze existing loss functions and introduce two novel perceptual losses—one CNN-based and one Transformer-based. Our approach improves recognition performance, achieving an average OCR accuracy of 85.14%.
format Article
id doaj-art-85d950c2dce44d03a10e08e89f349fd2
institution OA Journals
issn 2227-7390
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Mathematics
spelling doaj-art-85d950c2dce44d03a10e08e89f349fd22025-08-20T01:56:20ZengMDPI AGMathematics2227-73902025-05-011310167310.3390/math13101673A Deep Learning Framework of Super Resolution for License Plate Recognition in Surveillance SystemPei-Fen Tsai0Jia-Yin Shiu1Shyan-Ming Yuan2Institute of Computer Science and Engineering, Department of Electrical and Computer Engineering, National Yang Ming Chiao Tung University, Hsinchu Campus, Hsinchu 30010, TaiwanInstitute of Computer Science and Engineering, Department of Electrical and Computer Engineering, National Yang Ming Chiao Tung University, Hsinchu Campus, Hsinchu 30010, TaiwanInstitute of Computer Science and Engineering, Department of Electrical and Computer Engineering, National Yang Ming Chiao Tung University, Hsinchu Campus, Hsinchu 30010, TaiwanRecognizing low-resolution license plates from real-world scenes remains a challenging task. While deep learning-based super-resolution methods have been widely applied, most existing datasets rely on artificially degraded images, and common quality metrics poorly correlate with OCR accuracy. We construct a new paired low- and high-resolution license plate dataset from dashcam videos and propose a specialized super-resolution framework for license plate recognition. Only low-resolution images with OCR accuracy ≥5 are used to ensure sufficient feature information for effective perceptual learning. We analyze existing loss functions and introduce two novel perceptual losses—one CNN-based and one Transformer-based. Our approach improves recognition performance, achieving an average OCR accuracy of 85.14%.https://www.mdpi.com/2227-7390/13/10/1673license plate recognition (LPR)super resolution (SR)perceptual lossoptical character recognition (OCR)
spellingShingle Pei-Fen Tsai
Jia-Yin Shiu
Shyan-Ming Yuan
A Deep Learning Framework of Super Resolution for License Plate Recognition in Surveillance System
Mathematics
license plate recognition (LPR)
super resolution (SR)
perceptual loss
optical character recognition (OCR)
title A Deep Learning Framework of Super Resolution for License Plate Recognition in Surveillance System
title_full A Deep Learning Framework of Super Resolution for License Plate Recognition in Surveillance System
title_fullStr A Deep Learning Framework of Super Resolution for License Plate Recognition in Surveillance System
title_full_unstemmed A Deep Learning Framework of Super Resolution for License Plate Recognition in Surveillance System
title_short A Deep Learning Framework of Super Resolution for License Plate Recognition in Surveillance System
title_sort deep learning framework of super resolution for license plate recognition in surveillance system
topic license plate recognition (LPR)
super resolution (SR)
perceptual loss
optical character recognition (OCR)
url https://www.mdpi.com/2227-7390/13/10/1673
work_keys_str_mv AT peifentsai adeeplearningframeworkofsuperresolutionforlicenseplaterecognitioninsurveillancesystem
AT jiayinshiu adeeplearningframeworkofsuperresolutionforlicenseplaterecognitioninsurveillancesystem
AT shyanmingyuan adeeplearningframeworkofsuperresolutionforlicenseplaterecognitioninsurveillancesystem
AT peifentsai deeplearningframeworkofsuperresolutionforlicenseplaterecognitioninsurveillancesystem
AT jiayinshiu deeplearningframeworkofsuperresolutionforlicenseplaterecognitioninsurveillancesystem
AT shyanmingyuan deeplearningframeworkofsuperresolutionforlicenseplaterecognitioninsurveillancesystem