Monitoring and Detection of Tampered Number Plate Recognition

Abstract

The growing incidents of tampered and damaged vehicle number plates present significant challenges to law enforcement and intelligent transportation systems, as these alterations allow offenders to evade identification and accountability.This project, Detection and Monitoring Defective Number Plate System, aims to enhance traditional ANPR systems by incorporating automated detection and classification of tampered and damaged plates, offering a more comprehensive solution for traffic law enforcement and public safety. The system utilizes a combination of deep learning models, including YOLOv3 for real-time number plate detection and a Convolutional Neural Network (CNN) for classifying plates as tampered, damaged, or intact.

Introduction

I. Introduction

With the growing number of vehicles globally, monitoring and managing them efficiently is crucial. Manual tracking is becoming obsolete due to its limitations in accuracy and scalability. Automated Number Plate Recognition (ANPR) systems have emerged as a solution, utilizing computer vision to identify vehicle number plates from video feeds. These systems employ techniques like number plate extraction, segmentation, and character recognition to automate vehicle identification, enhancing accuracy and efficiency in various applications such as traffic monitoring, toll collection, and security enforcement.

II. Problem Statement

Vehicle number plates can become damaged or tampered over time due to environmental factors or intentional alterations. Such conditions pose challenges for ANPR systems, leading to:

1. Real-Time Detection: Identifying tampered or damaged plates promptly from video footage.

2. Classification: Accurately categorizing plates as tampered, damaged, or undamaged.

3. Character Recognition: Extracting readable characters from impaired plates.

Addressing these challenges is essential for maintaining the reliability and effectiveness of ANPR systems.

III. Literature Review

A. Intelligent System Using Convolutional Neural Networks (CNNs)

• Methodology:

  • Image Acquisition: Capture real-time images, converting them to grayscale for processing.

  • Super-Resolution: Enhance low-resolution images to improve recognition accuracy.

  • Segmentation: Apply bounding box methods to isolate individual characters.

  • Character Recognition: Utilize CNNs, specifically AlexNet, for character classification.en.wikipedia.org+1link.springer.com+1mdpi.com

• Key Findings:

  • Achieved 98.2% accuracy in number plate detection and 90.9% in character recognition.

  • Performed well under varying lighting and angle conditions.

  • Effective with low-resolution images.

• Strengths:

  • High accuracy and robustness.

  • Adaptability to different environmental conditions.

• Weaknesses:

  • High computational requirements.

  • Performance may degrade in real-time, high-speed scenarios.

B. Automatic License Plate Recognition Using Extracted Features

• Methodology:

  • Vertical Edge Detection Algorithm (VEDA): Detect vertical edges to identify number plate regions.

  • Image Normalization and Enhancement: Convert to grayscale and enhance image quality.

  • Segmentation: Use binary image transformation and connected component analysis.

  • Character Recognition: Implement template matching-based Optical Character Recognition (OCR).

• Key Findings:

  • Achieved 84.8% accuracy on a dataset of 500 real-world images.

  • Execution time under 0.5 seconds, suitable for real-time applications.

• Strengths:

  • Fast processing time.

  • Effective under various lighting conditions.

• Weaknesses:

  • Lower accuracy compared to CNN-based methods.

  • Limited robustness to tilted or distorted plates.en.wikipedia.org+2en.wikipedia.org+2en.wikipedia.org+2

IV. Proposed Solution

To enhance the detection and recognition of tampered or damaged number plates, the following integrated approach is proposed:

1. YOLOv7 for Real-Time Detection: Utilize YOLOv7 for fast and accurate detection of number plates in video streams.

2. CNN-Based Classification: Train a CNN to classify plates as tampered, damaged, or undamaged.

3. Optical Character Recognition (OCR): Implement OCR to extract characters from number plates, even if partially damaged.

4. Real-Time Monitoring and Alert System: Develop a system to provide immediate alerts upon detecting tampered or damaged plates.

This comprehensive system aims to improve the reliability and effectiveness of ANPR in various applications.

V. Methodology

The proposed system follows a structured approach:

1. Frame Extraction: Extract individual frames from video clips for processing.

2. Preprocessing:

   • Convert images to grayscale.

   • Enhance contrast using morphological operations.

   • Apply Gaussian filtering to reduce noise.

3. Thresholding: Use adaptive thresholding to separate the number plate from the background.

4. Edge Detection: Implement Canny edge detection to identify plate boundaries.

5. Plate Localization: Determine the location of the number plate using contour analysis.

This methodology ensures accurate and efficient detection and recognition of number plates.

VI. Results

The proposed system is expected to demonstrate high accuracy in detecting and recognizing number plates, including those that are tampered or damaged. Performance metrics such as detection accuracy, recognition rate, and processing time will be evaluated to assess the system's effectiveness.

VII. Advantages

• High Accuracy: Achieves high detection and recognition rates.

• Real-Time Processing: Capable of processing video streams in real-time.

• Robustness: Performs well under various environmental conditions.

• Scalability: Suitable for deployment in diverse applications.

VIII. Limitations

• Computational Requirements: High processing power needed for real-time operation.

• Environmental Factors: Performance may be affected by extreme conditions such as heavy rain or fog.

• Data Dependency: Requires a large and diverse dataset for training to ensure robustness.

Conclusion

In this research, a system is proposed for detecting and recognizing vehicle numberplates in Bangladesh, which are written in the Bengali language. In this system, the imagesof the vehicles are captured and then the number plate regions are extracted using the templatematching method. Then, the segmentation of each character is performed. Finally,a convolutional neural networks (CNN) is used for extracting features of each characterthat classifies the vehicle city, type, and number, to recognize the characters of the numberplate. The CNN provides a large number of features to help with accurate recognitionof characters from the number plate. This research used super resolution techniques torecognize characters with high resolution. In order to evaluate the experiment results,700 vehicle images were appointed. After training,the CNN acquired 98.2% accuracy based on the validation set, and attained 98.1% accuracybased on the testing set. This system can also be used for the number plates writtenin other languages in the same way.

References

[1] Different Types of Number Plates in India. Available online at https://www.godigit.com/traffic-rules/different-types-of-number-plates Last accessed on 17 August 2024 [2] Car Number Plate in India: A definitive Guide to RTO Number Plate. Available online at https://www.turtlemint.com/vehicle-number-plate/ Last accessed on 17 August 2024 [3] Statutory Plates and Inscriptions for Motor Vehicles, their Location and Method of attachment – Vehicle Identification Numbering System Ministry of Road Transport & Highways, Government Of India, 18th February 2024. Available online at https://morth.nic.in/sites/default/files/ASI/531201714455PMAIS-065_and_Amd_1_to_3.pdf Last accessed on 17 August 2024 [4] Nick Efford. “Morphological image processing”, Chapter 11, Digital Image Processing: A Practical Introduction Using JavaTM. Pearson Education, 2023. [Online]. Available: https://www.cs.auckland.ac.nz/courses/compsci723s1c/lectures/ImageProcessing-html/topic4.htm [5] Shashidhar, R., et al. \"Vehicle Number Plate Detection and Recognition using YOLO-v7 and OCR Method.\" 2023 IEEE International Conference on Mobile Networks and Wireless Communications (ICMNWC). IEEE, 2023. [6] Al Awaimri, Mohammed, et al. \"Vehicles Number Plate Recognition Systems A Systematic Review.\" 2023 International Conference on Computer, Control, Electrical, and Electronics Engineering (ICCCEEE). IEEE, 2023. [7] Maheswari, V. Uma, Rajanikanth Aluvalu, and Swapna Mudrakola. \"An integrated number plate recognition system through images using threshold-based methods and KNN.\" 2022 International Conference on Decision Aid Sciences and Applications (DASA). IEEE, 2022. [8] Mir, Md Nazmul Hossain, et al. \"IoT based digital toll collection system: A perspective.\" 2023 International Conference on Artificial Intelligence and Smart Systems (ICAIS). IEEE, 2024. [9] Lubna, Naveed Mufti, and Syed Afaq Ali Shah. \"Automatic number plate Recognition: A detailed survey of relevant algorithms.\" Sensors 21.9 (2024): 3028. [10] Rajput, Sudhir Kumar, et al. \"Automatic Vehicle Identification and Classification Model Using the YOLOv7 Algorithm for a Toll Management System.\" Sustainability 14.15 (2023): 9163. [11] Alam, Nur-A., et al. \"Intelligent system for vehicles number plate detection and recognition using convolutional neural networks.\" Technologies 9.1 (2024): 9. [12] Ahmed, Ahmed Abdelmoamen, and Sheikh Ahmed. \"A Real-Time Car Towing Management System Using ML-Powered Automatic Number Plate Recognition.\" Algorithms 14.11 (2024): 317. [13] Kumar, JMSV Ravi, B. Sujatha, and N. Leelavathi. \"Automatic vehicle number plate recognition system using machine learning.\" IOP Conference Series: Materials Science and Engineering. Vol. 1074. No. 1. IOP Publishing, 2024.

Copyright

Copyright © 2025 Shreyas MS, Yamini Rathore, Vivek Kumar, Vanukuri Prasant Reddy. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.