Related Experiment Video
Updated: Oct 30, 2025
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
728
Robust Korean License Plate Recognition Based on Deep Neural Networks
Hanxiang Wang1, Yanfen Li1, L-Minh Dang1
1Department of Computer Science and Engineering, Sejong University, Seoul 143-747(05006), Korea.
Sensors (Basel, Switzerland)
|July 2, 2021
Summary
This study presents a novel deep learning-based Korean License Plate Recognition (LPR) system. The robust LPR framework achieves 98.94% accuracy, effectively handling diverse environmental conditions and new license plate formats.
Area of Science:
- Computer Vision
- Artificial Intelligence
- Machine Learning
Background:
- License Plate Recognition (LPR) systems are crucial for vehicle management.
- Existing LPR systems struggle with new license plate formats and environmental variations.
Purpose of the Study:
- To develop a novel deep learning-based Korean LPR system.
- To address challenges posed by environmental conditions and evolving license plate designs.
Main Methods:
- Integration of three pre-processing techniques: defogging, low-light enhancement, and super-resolution.
- Development of two Korean LPR approaches: whole license plate recognition (W-LPR) and single-character license plate recognition (SC-LPR).
- Creation of synthetic and real-world Korean LPR datasets.
Main Results:
- The proposed LPR framework achieved a highest recognition accuracy of 98.94%.
- The system demonstrated robustness in recognizing license plates under various challenging conditions.
- Effectively handled a new type of Korean license plate.
Conclusions:
- The novel deep learning-based Korean LPR system offers a robust solution for effective vehicle management.
- The integrated pre-processing techniques and specialized recognition approaches enhance LPR performance.
- The developed datasets support further research and development in Korean LPR.
