Why a Smarter Template Matching Method Could Improve License Plate Detection in Intelligent Transportation Systems

on

Automatic license plate detection has become one of the fundamental technologies supporting intelligent transportation systems (ITS), smart cities, electronic toll collection, traffic law enforcement, and vehicle access management. Reliable localization of license plates is essential because it serves as the first stage of license plate recognition (LPR), directly affecting the accuracy of subsequent character segmentation and recognition processes. Although deep learning has significantly improved detection performance in recent years, many practical applications still require computationally efficient methods that can operate with limited processing resources and without extensive training datasets. This study introduces an adaptive Euler-template matching method that seeks to improve the efficiency of template-based license plate localization while maintaining competitive detection accuracy. By proposing a novel template representation and an adaptive matching strategy, the research offers an alternative approach for fast and practical license plate detection in real-world intelligent transportation applications.

Bibliographic Information

Item Information
Article Title Localization of Car License Plate Using Adaptive Euler-Template Matching Method
Authors Nay Zar Aung, Jinghui Peng, Khin Cho Tun, and Songjing Li
Journal Innovation in Engineering
Volume & Issue Volume 2, Issue 2
Publication Year 2025
Pages 91–105
DOI https://doi.org/10.58712/ie.v2i1.30
Publisher Researcher and Lecturer Society
License Creative Commons Attribution 4.0 International (CC BY 4.0)

1. Research Background

  • License plate localization is the foundation of automatic license plate recognition (LPR). Before characters can be segmented and recognized, the system must first accurately identify the position of the license plate within the vehicle image. Errors at this stage propagate throughout the entire recognition process.
  • Intelligent transportation systems increasingly depend on reliable license plate detection. Modern smart traffic control systems integrate cameras, sensors, processing units, and communication networks to automate vehicle monitoring, toll collection, traffic management, and law enforcement. Accurate license plate localization is therefore a critical enabling technology.
  • Detecting license plates remains technically challenging. License plates often resemble other rectangular objects found on vehicles, including headlights, mirrors, radiators, and body components. In outdoor environments, windows, signboards, brick patterns, and other background objects may further complicate detection.
  • Existing detection approaches involve different trade-offs. Previous studies have explored edge-based methods, color-based segmentation, region analysis, wavelet transforms, and more recently deep learning techniques. While deep neural networks often achieve high detection accuracy, they generally require large annotated datasets and extensive training procedures.
  • Template matching remains attractive because of its simplicity. Conventional template matching algorithms are relatively easy to implement and require limited computational complexity compared with many learning-based methods. However, traditional approaches often suffer from high computational cost, sensitivity to pixel-level misalignment, and dependence on manually designed templates.
  • Computational efficiency is an important practical consideration. Many real-world intelligent transportation applications require algorithms capable of operating quickly on standard computing hardware while maintaining reliable detection performance under varying image conditions.
  • The study introduces a new Euler-template concept. Instead of relying on conventional gray-level templates, the proposed approach constructs binary templates based on Euler number characteristics, enabling adaptive template resizing and reducing unnecessary template searching across the entire image.
  • The research also proposes a new template matching strategy. Rather than comparing individual pixels directly, the method evaluates regional average similarities, making template matching less sensitive to slight positional differences between the template and target image.
  • The primary contribution is an adaptive template matching framework for license plate localization. The proposed methodology combines Euler-number filtering, adaptive template generation, candidate region selection, and a novel similarity calculation to improve both computational efficiency and detection robustness without requiring model training.

2. Research Objectives

  • To develop an adaptive Euler-template matching method for automatic localization of vehicle license plates.
  • To introduce a new binary template representation based on Euler-number characteristics for license plate detection.
  • To design two different Euler-template models and compare their detection performance.
  • To propose a new similarity measurement concept that reduces sensitivity to pixel-level mismatches during template matching.
  • To reduce the computational cost of conventional template matching by performing adaptive matching only within candidate regions instead of scanning the entire image.
  • To experimentally evaluate the proposed method using multiple vehicle images and compare its accuracy and processing time with several existing license plate localization approaches.

3. Why This Research Matters

  • Supports intelligent transportation systems. Fast and reliable license plate localization contributes directly to vehicle identification, automated toll collection, traffic monitoring, parking management, and smart mobility applications.
  • Provides a computationally efficient alternative to deep learning. The proposed method avoids extensive model training while still achieving competitive detection performance, making it attractive for applications with limited computational resources.
  • Reduces processing time. By filtering candidate regions before template matching and adaptively resizing templates, the algorithm minimizes unnecessary computations commonly associated with traditional template matching.
  • Improves robustness against minor image variations. The regional-average matching strategy decreases the influence of small pixel displacements, allowing more stable detection under practical imaging conditions.
  • Demonstrates an innovative application of Euler-number concepts. The introduction of Euler-template models illustrates how topological image properties can be incorporated into practical computer vision algorithms for object localization.
  • Facilitates engineering implementation. Because the methodology is relatively straightforward and does not depend on large training datasets, it can be implemented more easily in embedded vision systems, industrial inspection platforms, and intelligent transportation devices.
  • Offers broader applicability. Although developed for vehicle license plate localization, the adaptive template matching concept may inspire similar approaches for detecting other structured objects with consistent geometric characteristics in machine vision applications.

4. Research Methodology

  • Research Type

    This study employed an experimental computer vision approach to develop and evaluate a new template-based algorithm for automatic license plate localization. Rather than adopting machine learning or deep neural networks, the researchers proposed a deterministic image-processing framework that combines candidate region filtering, adaptive template generation, and template matching to identify license plates efficiently.

  • Research Framework

    The proposed framework consists of four principal stages: image pre-processing, candidate region selection, adaptive Euler-template matching, and license plate localization. The workflow begins with image acquisition, followed by grayscale conversion and binary image generation. Image regions satisfying predefined geometric and topological conditions are retained as candidate regions before adaptive template matching is performed to determine the final license plate location.

  • Image Acquisition and Pre-processing

    Vehicle images were initially captured as RGB color images before being converted into grayscale using a customized grayscale transformation designed to emphasize license plate characters while minimizing the influence of different plate background colors. The grayscale image was subsequently transformed into a binary image through thresholding, enabling the calculation of Euler numbers for each connected region.

  • Euler Number Concept

    The proposed method utilizes the Euler number as a topological descriptor of binary images. Euler number is determined from the relationship between connected foreground objects and internal holes. Because license plates contain multiple character regions represented as holes in the binary image, Euler numbers provide an effective criterion for distinguishing potential license plate regions from other rectangular objects appearing in vehicle images.

  • Development of Euler-Templates

    Two different binary template models were designed for the study. Euler-template Model-A represents license plate characters using rectangular holes that approximate character positions under lower image quality or geometric distortion. Euler-template Model-B employs digit-shaped holes that more closely resemble actual character structures and is intended for images with clearer character representation.

    Unlike conventional templates, both Euler-template models are generated using binary patterns and can be resized proportionally to match candidate regions without requiring multiple predefined template sizes or color variations.

  • Candidate Region Selection

    To reduce computational complexity, the algorithm performs template matching only within selected candidate regions rather than across the entire image. Candidate regions are filtered using two criteria: Euler number and aspect ratio. Regions satisfying the predefined thresholds are retained as potential license plate locations, while non-relevant objects are discarded before template matching begins.

  • Adaptive Euler-Template Matching

    Instead of shifting a fixed template across every image location, the proposed method adaptively resizes the Euler-template to match each candidate region individually. This strategy substantially reduces unnecessary computations while allowing the template to accommodate different license plate dimensions.

    To improve robustness, similarity is evaluated using regional average values rather than strict pixel-by-pixel correspondence. This regional comparison minimizes errors caused by small positional shifts between the template and the target object.

  • Experimental Dataset

    Performance evaluation was conducted using 150 vehicle images containing Myanmar license plates. The same dataset was used to evaluate both Euler-template models. An additional experiment employed Chinese family-car license plates to demonstrate the adaptability of the proposed approach to a different license plate format through template redesign.

  • Software and Computing Environment

    The algorithm was implemented using MATLAB 2014 on a computer equipped with an Intel® Core™ i5-7500 processor operating at 3.40 GHz and 8 GB RAM. Performance was assessed using precision and average processing time recorded during experimental execution.

  • Validation and Performance Evaluation

    The proposed method was validated by comparing its detection accuracy and processing time with several previously published license plate localization methods. Experimental evaluation examined the performance of both Euler-template models, their robustness under varying vehicle appearances and viewing perspectives, and their applicability to different license plate formats.

5. Key Findings

Adaptive Euler-Template Matching Achieved High Detection Accuracy

The experimental evaluation demonstrates that the proposed adaptive Euler-template matching method can localize vehicle license plates with high precision while maintaining relatively low computational cost. Among the two proposed templates, Euler-template Model-B achieved the best overall performance, correctly detecting license plates in 145 of the 150 test images, corresponding to a precision of 96.7%. Euler-template Model-A also produced competitive results with a precision of 96.0%, indicating that both template designs are effective for practical license plate localization.

These results show that carefully designed template-based methods remain competitive despite the increasing popularity of deep learning approaches. The proposed algorithm demonstrates that robust detection can be achieved without extensive model training or very large image datasets.

Adaptive Candidate Region Selection Significantly Reduced Computational Work

A major contribution of the proposed framework is the elimination of exhaustive template searching across the entire image. Instead of sliding templates pixel by pixel, the algorithm first identifies potential candidate regions using Euler number and aspect ratio constraints before performing adaptive template matching only within those selected regions.

This strategy substantially decreases unnecessary computations while preserving detection reliability. The average processing time for both Euler-template models was approximately 0.303 seconds per image, demonstrating the computational efficiency of the proposed localization framework.

Regional Similarity Matching Improved Robustness Against Pixel Misalignment

Traditional template matching algorithms often suffer from significant matching errors when small positional shifts occur between template and target images. To overcome this limitation, the proposed method introduces a regional averaging strategy that evaluates local neighborhoods rather than relying exclusively on exact pixel correspondence.

This modification makes the similarity calculation more tolerant of slight geometric variations and minor alignment errors. Consequently, the proposed algorithm maintains reliable detection performance even when perfect pixel alignment cannot be achieved during template matching.

Euler-Template Model-B Outperformed the Simpler Template Design

Comparative experiments reveal that Euler-template Model-B consistently produced lower Sum of Squared Difference (SSD) values than Model-A across representative test images. Because Model-B incorporates digit-shaped holes that more closely resemble actual license plate characters, it provides a better structural representation of license plates during template matching.

Although Model-A remains useful for lower-quality or distorted images, the experimental results indicate that Model-B offers greater matching accuracy when character structures are sufficiently visible within the captured image.

The Algorithm Demonstrated Robustness Under Diverse Imaging Conditions

Experimental examples show that the proposed method successfully detected license plates across different vehicle colors, outdoor environments, and complex backgrounds. Multiple non-license plate objects—including mirrors, headlights, radiators, and surrounding structures—were initially identified as candidate regions but were subsequently rejected during template matching.

The method also remained effective when license plates appeared with moderate perspective distortion. Experimental observations indicated successful localization for viewing angles approaching approximately 15°, demonstrating a degree of robustness to practical imaging conditions encountered in traffic surveillance systems.

The Proposed Framework Can Be Adapted to Different License Plate Formats

Beyond Myanmar license plates, the researchers redesigned Euler-template Model-B for Chinese family-car license plates by modifying the template geometry according to the different aspect ratio and character arrangement. Experimental demonstrations confirmed that the adaptive template concept remained applicable after redesigning the template for the new plate format.

This adaptability suggests that the proposed methodology is not restricted to a single national license plate design. Instead, the Euler-template concept provides a flexible framework that can be customized for different license plate standards without fundamentally changing the overall detection algorithm.

The Proposed Method Provides a Competitive Balance Between Accuracy and Efficiency

Comparison with previously published license plate localization techniques shows that the proposed method achieves detection accuracy comparable to several established approaches while requiring relatively short processing time. Although some existing algorithms reported marginally higher detection rates, they generally involved greater algorithmic complexity or longer execution times.

The study therefore demonstrates that adaptive Euler-template matching offers an effective compromise between computational efficiency and localization performance, making it particularly attractive for engineering applications where rapid processing and implementation simplicity are important design considerations.

6. Scientific Contribution

  • Introduces the concept of Euler-Template for license plate localization. The study proposes a new binary template representation based on Euler-number characteristics, extending the application of topological image descriptors from image analysis to practical object localization in intelligent transportation systems.
  • Develops an adaptive template matching framework. Instead of performing exhaustive template searching across the entire image, the proposed method adaptively resizes the template according to each candidate region, significantly reducing computational effort while preserving detection performance.
  • Proposes a new similarity calculation strategy. The research introduces regional average matching as an alternative to conventional pixel-by-pixel comparison, making template matching more tolerant of small positional shifts and improving robustness against minor image variations.
  • Demonstrates two complementary Euler-template designs. The comparison between Model-A and Model-B provides engineering insight into how template geometry influences localization performance under different image quality conditions, with Model-B achieving the highest experimental accuracy.
  • Integrates topological filtering with adaptive template matching. By combining Euler-number analysis, aspect-ratio filtering, adaptive template generation, and regional similarity measurement, the study presents a coherent and computationally efficient license plate localization framework.
  • Provides a practical alternative to data-intensive learning approaches. The proposed method demonstrates that high localization performance can still be achieved through carefully designed deterministic image-processing techniques without requiring extensive training datasets or complex neural network architectures.
  • Demonstrates adaptability across different license plate formats. The successful redesign of the Euler-template for Chinese license plates illustrates that the methodology can be customized for different national plate standards while preserving the overall detection framework.

7. Industrial Implications

  • Supports intelligent transportation infrastructure. Reliable and computationally efficient license plate localization can improve vehicle identification systems used in smart traffic control, electronic toll collection, parking automation, and intelligent transportation networks.
  • Facilitates embedded vision applications. Because the proposed algorithm does not require computationally expensive model training, it is suitable for deployment on embedded processors, edge-computing devices, and resource-constrained traffic monitoring equipment.
  • Improves real-time traffic monitoring. Reduced processing time enables faster vehicle detection, allowing traffic management systems to process larger numbers of vehicle images with lower computational requirements.
  • Enhances automatic vehicle access systems. The methodology may support automated gate control, restricted-area access, parking management, and secure vehicle authentication where dependable license plate localization is required before recognition.
  • Provides flexibility for different regional license plate standards. The adaptive Euler-template concept allows engineers to redesign templates according to local license plate layouts without modifying the overall localization algorithm, simplifying implementation across multiple countries or jurisdictions.
  • Supports cost-effective engineering implementation. Organizations that lack access to high-performance computing resources or large annotated datasets may benefit from a deterministic computer vision approach that remains relatively straightforward to implement and maintain.
  • Contributes to digital engineering applications. Efficient image localization algorithms such as the proposed method can be integrated into broader Industry 4.0 and smart city ecosystems where automated image analysis supports digital infrastructure management and intelligent transportation services.

8. Research Limitations

  • The experimental evaluation primarily used 150 vehicle images containing Myanmar license plates. Although the dataset enabled comprehensive performance assessment of the proposed algorithm, additional validation using larger and more diverse image collections would provide broader evidence of general applicability.
  • The proposed Euler-template models were specifically designed according to the structural characteristics of Myanmar license plates, with a separate template subsequently developed for Chinese license plates. Other national license plate formats would require corresponding template redesign before implementation.
  • The algorithm depends on successful candidate region selection based on Euler number and aspect ratio. If the license plate region merges with neighboring objects or cannot satisfy these filtering criteria, localization performance may decrease.
  • Experimental observations indicate that the method remains effective under moderate perspective distortion, approximately up to 15 degrees. More severe viewpoint changes or extreme geometric distortions were not investigated in the present study.
  • The research focused exclusively on the license plate localization stage of the overall license plate recognition process. Character segmentation, optical character recognition, and complete end-to-end license plate recognition performance were beyond the scope of this investigation.
  • Performance comparisons with previous studies were based on reported accuracy and processing time. Because experimental datasets, hardware configurations, and implementation environments differ among published studies, direct comparisons should be interpreted with appropriate caution.

9. Future Research Opportunities

  • Evaluate the proposed adaptive Euler-template matching method using substantially larger and more diverse international license plate datasets collected under different environmental conditions.
  • Investigate the robustness of the algorithm under more challenging imaging scenarios, including severe perspective distortion, partial occlusion, motion blur, nighttime imaging, adverse weather conditions, and varying illumination levels.
  • Develop automated procedures for generating Euler-templates from different national license plate standards, reducing the need for manual template design.
  • Combine the proposed localization framework with modern character segmentation and optical character recognition techniques to establish a complete end-to-end license plate recognition system.
  • Explore hybrid approaches that integrate adaptive Euler-template matching with machine learning or deep learning methods to balance computational efficiency and detection accuracy.
  • Investigate adaptive threshold selection strategies for candidate region filtering to improve performance across wider variations in image quality and environmental conditions.
  • Implement the algorithm on embedded hardware, edge-computing platforms, or field-programmable gate arrays (FPGAs) to evaluate real-time deployment capability in intelligent transportation systems.
  • Compare the proposed regional similarity measurement with alternative similarity metrics to determine their influence on localization accuracy and computational efficiency.
  • Extend the Euler-template concept to other computer vision applications involving structured objects with consistent geometric characteristics, such as industrial inspection, document analysis, or symbol recognition.
  • Assess the long-term operational performance of the proposed method within real intelligent transportation infrastructures, including smart parking systems, electronic toll collection, and automated traffic enforcement applications.

10. Potential for Public Policy Citation (Overton)

This article has moderate potential to be cited in public policy documents because it addresses a practical engineering challenge associated with intelligent transportation systems and smart mobility infrastructure. The proposed adaptive Euler-template matching method offers a computationally efficient solution for automatic license plate localization, which is an essential component of vehicle identification technologies used in modern transportation management.

The methodology may serve as a useful technical reference for transportation authorities, highway agencies, and organizations responsible for developing intelligent traffic management systems. Applications such as electronic toll collection, automated parking management, traffic surveillance, vehicle access control, and smart city infrastructure all depend on reliable license plate localization techniques. Consequently, the study may contribute to technical guidance documents, intelligent transportation implementation strategies, and digital infrastructure development initiatives.

The article also aligns with broader digital engineering and Industry 4.0 initiatives by demonstrating how lightweight computer vision algorithms can support automated transportation services without relying on computationally intensive deep learning models. Such characteristics may increase its relevance for engineering guidelines that emphasize efficient deployment on embedded or resource-constrained hardware platforms.

Nevertheless, the policy relevance remains primarily technical rather than regulatory. The research focuses on algorithm development and experimental validation instead of evaluating transportation governance, legal frameworks, economic impacts, or nationwide implementation strategies. Therefore, the article is more likely to be referenced in engineering manuals, intelligent transportation technology roadmaps, technical standards, and digital infrastructure guidance than in broader public policy documents.

11. Who Should Read This Paper?

  • Researchers working in computer vision and intelligent transportation systems.
  • Engineers developing automatic license plate recognition (ALPR/LPR) systems.
  • Graduate students in computer engineering, electrical engineering, and artificial intelligence.
  • Researchers studying image processing, pattern recognition, and machine vision.
  • Transportation engineers involved in smart traffic control systems.
  • Developers of embedded vision and edge-computing applications.
  • Industry practitioners working on intelligent surveillance and vehicle monitoring technologies.
  • Smart city planners implementing intelligent transportation infrastructure.
  • Government agencies responsible for transportation technology deployment.
  • Educators teaching digital image processing, intelligent systems, and computer vision.

12. Final Thoughts

This study presents a thoughtful and practical contribution to the field of automatic license plate localization by revisiting template matching from a new perspective. Rather than competing directly with increasingly complex deep learning approaches, the authors demonstrate that carefully designed deterministic image-processing techniques can still deliver strong localization performance while maintaining computational efficiency. The introduction of Euler-template models, adaptive template resizing, and regional similarity matching collectively provides an innovative framework that addresses several well-known limitations of conventional template matching.

One of the major strengths of the research lies in its balance between simplicity and engineering effectiveness. By filtering candidate regions using Euler-number characteristics before adaptive matching, the proposed algorithm significantly reduces unnecessary computations while preserving high localization accuracy. The comparative evaluation of two template models further enriches the study by illustrating how template geometry influences detection performance under different imaging conditions. The successful adaptation of the methodology to Chinese license plates also demonstrates the flexibility of the overall framework.

Although additional validation using larger datasets and more challenging imaging scenarios would strengthen confidence in the algorithm's broader applicability, the study clearly demonstrates the continued relevance of efficient template-based computer vision methods. For intelligent transportation systems that require reliable, fast, and resource-efficient license plate localization, the adaptive Euler-template matching method offers a valuable alternative to computationally intensive learning-based approaches. Overall, this research makes a meaningful engineering contribution by combining topological image analysis with adaptive template matching into a practical solution for modern vehicle identification systems.

Suggested Citation

UNP–Teknomekanik Style

Aung NZ, Peng J, Tun KC, Li S. Localization of car license plate using adaptive Euler-template matching method. Innovation in Engineering. 2025;2(2):91–105. Available from: https://doi.org/10.58712/ie.v2i1.30.

APA (7th Edition)

Aung, N. Z., Peng, J., Tun, K. C., & Li, S. (2025). Localization of car license plate using adaptive Euler-template matching method. Innovation in Engineering, 2(2), 91–105. https://doi.org/10.58712/ie.v2i1.30

IEEE Style

N. Z. Aung, J. Peng, K. C. Tun, and S. Li, "Localization of car license plate using adaptive Euler-template matching method," Innovation in Engineering, vol. 2, no. 2, pp. 91–105, 2025, doi: 10.58712/ie.v2i1.30.

Harvard Style

Aung, N.Z., Peng, J., Tun, K.C. and Li, S., 2025. Localization of car license plate using adaptive Euler-template matching method. Innovation in Engineering, 2(2), pp.91–105. Available at: <https://doi.org/10.58712/ie.v2i1.30>.

Vancouver Style

Aung NZ, Peng J, Tun KC, Li S. Localization of car license plate using adaptive Euler-template matching method. Innovation in Engineering. 2025;2(2):91–105. doi: 10.58712/ie.v2i1.30.

Chicago (Author–Date)

Aung, Nay Zar, Jinghui Peng, Khin Cho Tun, and Songjing Li. 2025. "Localization of Car License Plate Using Adaptive Euler-Template Matching Method." Innovation in Engineering 2 (2): 91–105. https://doi.org/10.58712/ie.v2i1.30.

MLA (9th Edition)

Aung, Nay Zar, et al. "Localization of Car License Plate Using Adaptive Euler-Template Matching Method." Innovation in Engineering, vol. 2, no. 2, 2025, pp. 91–105. https://doi.org/10.58712/ie.v2i1.30.

Editorial Note

Editorial Note: This blog post is an independent scholarly review intended for educational and scientific communication purposes. It summarizes and discusses the published article in the author's own words while providing full attribution to the original publication, consistent with the principles of the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

SEO Information

SEO Meta Description

A comprehensive scholarly review of the adaptive Euler-template matching method for automatic license plate localization. Discover how this research improves computational efficiency, template matching, and intelligent transportation systems without relying on deep learning.

SEO Keywords

Adaptive Euler-template, License Plate Localization, Automatic License Plate Recognition, ALPR, License Plate Detection, Template Matching, Computer Vision, Intelligent Transportation Systems, Smart Traffic Control, Image Processing, Euler Number, Pattern Recognition, Machine Vision, MATLAB, Embedded Vision, Vehicle Detection, Digital Engineering, Smart City, Engineering Research Review, Innovation in Engineering.

Recommended URL Slug

adaptive-euler-template-license-plate-localization-review

Discoverability

  • Computer Vision
  • Image Processing
  • Pattern Recognition
  • Automatic License Plate Recognition (ALPR)
  • License Plate Detection
  • Intelligent Transportation Systems (ITS)
  • Smart Traffic Control
  • Smart Cities
  • Machine Vision
  • Digital Engineering
  • Embedded Systems
  • Industry 4.0
  • Transportation Engineering
  • Engineering Research Insights
  • Innovation in Engineering Journal

Engineering Research Insights — Independent Scholarly Article Review

Comments

Post a Comment