International Journal of Innovative Research in Computer Science and Technology (IJIRCST)

ISSN (Online): 2347-5552, Volume-12, Issue-5 September 2024

https://doi.org/10.55524/ijircst.2024.12.5.5
Article ID IRP-1552, Pages 28-35
www.ijircst.org

Recent Developments in Automatic Number Plate Detection and

Recognition
Faraz Imtiyaz Mir1, Nibbritta Niloy Sarkar2, *Dr. Yojna Arora3, and Dr Avinash Kumar
Sharma4
1, 2
Student, Department of Computer Science Engineering, Sharda University, Greater Noida, Uttar Pradesh, India
3, 4
Associate Professor, Department of Computer Science Engineering, Sharda University, Greater Noida, Uttar Pradesh,
India
Correspondence should be addressed to Dr Yojna Arora;

Received 3 August 2024; Revised 18 August 2024; Accepted 2 September 2024

Copyright © 2024 made by Dr Yojna Arora et al . 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.

ABSTRACT- Automatic Number Plate Recognition various environments, from highways to residential areas.
(ANPR) is a mass surveillance system that captures the This capability is crucial for tracking stolen vehicles,
image of vehicles and recognizes their license numbers. managing traffic, and enhancing overall public safety [4]. A
ANPR can be used in the detection of stolen vehicles. The general architecture for number plate detection is mentioned
detection of stolen vehicles can be done in an efficient in below figure 1 [5].
manner by using the ANPR systems located in the
highways. This method presents a recognition method in
which the vehicle plate image is obtained by the digital
cameras and the image is processed to get the number plate
information. An image of a vehicle is captured and
processed using image pre-processing algorithm. In this
context, the number plate area is localized using edge
detection method and the characters acquired are segmented
using segmentation method. The segmented characters are
passed to Optical character reader (OCR) which gives exact
characters of the number plate and are stored in database.
The image pre-processing is done using Gaussian Blur
method, conversion of RGB (Red, Green, Blue) to
grayscale, Edge detection using sobel followed by contour
detection and license plate is extracted. OCR is run on the Figure 1: General Architecture for Number Plate Detection
Region of Interest (ROI) given by the contour analysis. The
OCR provides the vehicle’s num-ber obtained from the
number plate and is stored in the database which is used for This paper focuses on detailing the application of Faster R-
surveillance and tracking vehicle’s location. The software CNN for the development of an ANPR system, outlining
used in the process is Python version 3 and different python the technological framework and methodologies employed
libraries like Tesseract OCR, OpenCV. [6]. The objectives are twofold: firstly, to demonstrate how
digital image processing techniques can be utilized to detect
KEYWORDS-: Automatic Number Plate Recognition and read vehicle number plates [7] and secondly to
(ANPR), surveillance, image pre-processing, Gaussian Blur highlight the system's application in real-world scenarios,
method, Optical character reader (OCR). emphasizing its role in surveillance and the detection of
stolen vehicles [8].
I. INTRODUCTION
The advent of Automated Number Plate Recognition
II. LITERATURE REVIEW
(ANPR) systems represents a significant technological This section explores the core aspects of Automated
advance in surveillance and security measures for vehicular Number Plate Recognition (ANPR) systems, including their
monitoring [1]. Leveraging Faster R-CNN, a powerful deep conceptual framework, technological components, and
learning model, this chapter explores the implementation of operational workflow [11]. These systems play a pivotal
ANPR for effective vehicle identification and security [2]. role in vehicle surveillance and management, employing
Automatic Number Plate Recognition systems are integral advanced image processing and machine learning
to modern surveillance strategies, aiding significantly in technologies to enhance public safety and security.
law enforcement and security by automating the detection Several case studies highlight the effectiveness of applying
and recording of vehicle license plates[3]. The significance Faster R-CNN to ANPR systems:
of these systems lies in their ability to quickly and Urban Traffic Monitoring: In a city traffic system, Faster R-
accurately capture and process vehicular information across CNN-based ANPR was used to monitor and manage city
Innovative Research Publication 28
 International Journal of Innovative Research in Computer Science and Technology (IJIRCST)

traffic effectively, helping in law enforcement and traffic printed text, or natural imagery is known as optical
rule compliance [13]. character recognition, or OCR [16]. DIP is the procedure
Stolen Vehicle Detection: ANPR systems equipped with wherein a computer algorithm is used to manipulate digital
Faster R-CNN have been instrumental in identifying and photographs. DIP is a field that continues to advance with
recovering stolen vehicles by rapidly scanning and time and has applications in almost every other industry,
verifying license plate numbers against a database of including robotics, banking, healthcare, and PET sweeps
reported stolen vehicles [14]. [17]. Pattern recognition, which encompasses picture
Parking Management: Automated parking systems use recognition, handwriting recognition, and computer-aided
Faster R-CNN-based ANPR to manage entry and exit of diagnosis, is one of its main uses [18].
vehicles, ensuring non-intrusive and efficient parking
operations [15].
The computerized conversion of text or creation of a digital
copy of text from sources such as handwritten papers,
Table 1: Comparison of various algorithms used for Automatic Number Plate Detection
SNO. AUTHER AIM TECHNIQUES ADVANTAGES APPLICATION RESULT

1. Tote, A. S., The study's main goal is TensorFlow, High Accuracy, Traffic The system
Pardeshi, S. to use TensorFlow to OpticalCharacter Real-time Management, achieved a
S., & Patange create an automatic Recognition Processing, Law detection
[1] number plate detection (OCR), Adaptability Enforcement, accuracy of
system that is Image Automated Toll approximately
specifically suited for Processing Collection, 95%.
the Indian environment. Parking
The system reads and Management,
recognizes vehicle
number plates with
accuracy thanks to
optical character
recognition (OCR)
capabilities.
2. Shambharkar, To develop an Optical High accuracy Traffic Achieved a
Y., automatic framework Character and precision in management and detection
Salagrama, S., for number plate Recognition detection and enforcement accuracy of
Sharma, K., detection using OCR (OCR) recognition Automated toll approximately
Mishra, O., & and deep learning Deep Learning Robustness to collection systems 94%.
Parashar, techniques Convolutional various lighting Parking
D.[2] Neural conditions and management
Networks plate designs solutions
(CNN) Real-time
Image processing
Processing capabilities
Techniques
3. Jawale, M. A., To implement a number Internet of Integration with Vehicle Achieved a
William, P., plate detection system Things (IoT) IoT for real-time registration and recognition
Pawar, A. B., for vehicle registration Convolutional data transmission monitoring accuracy of
& Marriwala, using IoT and Neural High accuracy Automated toll approximately
N. [3] recognition using CNN. Networks and reliability collection 93%.
(CNN) Scalability for Traffic law
Image large-scale enforcement
Processing deployment
4. Mozumder, To develop a hybrid Hybrid Edge High precision in Traffic Achieved a
M., Biswas, edge algorithm for Detection edge detection monitoring and detection
S., vehicle license plate Algorithm and plate control accuracy of
Vijayakumari, detection. Image localization Automated approximately
L., Naresh, R., Processing Efficient parking systems 92%.
Kumar, C. V., Techniques processing time Security and
& Karthika, Machine Robust to varying surveillance
G.[4] Learning environmental
conditions
5. Khinchi, M., To review various Literature Comprehensive Guidance for Summarizes
& Agarwal, technologies and review of understanding of developing ANPR various studies
C.[5] methods used in existing ANPR ANPR systems and reports
automatic number plate methods technologies Reference for detection
recognition (ANPR). Analysis of Identification of researchers and accuracies
different OCR strengths and developers in the ranging from
techniques weaknesses of field 85% to 95%.
Evaluation of different methods
deep learning Provides insights

Innovative Research Publication 29

 International Journal of Innovative Research in Computer Science and Technology (IJIRCST)

approaches for future

research
directions

6. Sasi, S., To develop a system for Optical High accuracy in Traffic Achieved a
Sharma, S., & automatic car number Character character enforcement recognition
Cheeran, A. plate recognition. Recognition recognition Automated toll accuracy of
N. [6] (OCR) Real-time systems approximately
Image processing Parking 90%.
Processing capability management
Techniques Adaptability to
Machine different plate
Learning designs
Algorithms
7. Gautam, A., To utilize a deep Deep Learning High accuracy Traffic Achieved a
Rana, D., learning approach for Convolutional and reliability monitoring recognition
Aggarwal, S., automatic recognition of Neural Robust Automated toll accuracy of
Bhosle, S., & license number plates. Networks performance collection approximately
Sharma, H. (CNN) under various Parking 96%
[7] Image conditions management
Processing Scalability for systems
Techniques large-scale
deployments
8. Prajapati, R. To provide an in-depth Review of Comprehensive Reference for Summarizes
K., Bhardwaj, analysis of machine Machine understanding of researchers and various studies
Y., Jain, R. learning techniques in Learning ANPR developers with detection
K., & Hiran, automatic number plate Techniques technologies Insights for accuracies
K. K. [8] recognition (ANPR). Analysis of Identification of improving ANPR ranging from
OCR and Deep key opportunities systems 85% to 97%.
Learning and limitations
Approaches Provides
Evaluation of guidance for
Different future research
Models and
Algorithms
9. Mittal, R., & To systematically Systematic Comprehensive Reference for Summarizes
Garg, A. [9] review text extraction Review of OCR understanding of researchers and various studies
techniques using OCR Methods OCR methods developers in with text
Analysis of Text Identification of OCR extraction
Extraction strengths and Guidance for accuracies
Techniques weaknesses developing text ranging from
Evaluation of Insights for future extraction 80% to 95%
Different research systems
Approaches directions
10. Ali, F., To develop a license Optical High accuracy in Traffic Achieved a
Rathor, H., & plate recognition system Character license plate enforcement recognition
Akram, W. Recognition recognition Automated toll accuracy of
[10] (OCR) Real-time collection approximately
Image processing Parking 92%.
Processing capability management
Techniques Adaptability to
Machine different
Learning environmental
Algorithms conditions

III. METHODOLOGY  Technological Framework: The technological

backbone of an ANPR system comprises several key
The conceptual framework of an ANPR system revolves components:
around the integration of image capture and processing
technologies to recognize vehicle number plates  Digital Cameras: High-resolution cameras capture clear
automatically [9]. Utilizing cameras installed at strategic images of moving vehicles under various lighting and
locations, such as traffic lights and highway tolls, the weather conditions [13].
system captures real-time images of vehicles. These images  Image Pre-processing Tools: Techniques like Gaussian
are then processed using a series of algorithms designed to Blur and grayscale conversion help enhance image
detect, segment, and interpret the characters on the license quality and reduce noise, preparing the images for
plates [10]. (See figure 2 ) further analysis[15].

Innovative Research Publication 30

 International Journal of Innovative Research in Computer Science and Technology (IJIRCST)

 Optical Character Recognition (OCR): Tesseract OCR  Operational Workflow: The operational workflow of an
and other OCR tools are employed to extract ANPR system can be described in several steps:
alphanumeric characters from the processed images[16].  Image Acquisition: Vehicles are captured in real-time
 Machine Learning Models: Faster R-CNN (Region- using strategically placed cameras[20].
based Convolutional Neural Networks) is specifically  Image Pre-processing: Captured images undergo
utilized for its efficiency in detecting objects (in this various preprocessing methods such as noise reduction,
case, number plates) within images[17]. edge detection, and contour analysis to isolate the
 Databases and Software: The recognized plate numbers number plate [21].
are stored in databases, and software like Python and its  Plate Detection: Using Faster R-CNN, the system
libraries (OpenCV) are used for implementing the identifies the region of the image that likely contains the
algorithms [18]. number plate [19].

Figure 2: Proposed system model [1]

 Character Segmentation and Recognition: Characters enhance the quality and improve the accuracy of the
on the detected plates are segmented and then subsequent recognition processes [21]
recognized using OCR technology [21].  Gaussian Blur: Applied to smooth the image, reducing
 Data Storage and Usage: The extracted plate noise and details that are not required, which helps in
information is stored in a database, which can be enhancing the accuracy of edge detection [25].
accessed for surveillance purposes, such as tracking  Grayscale Conversion: Converting images from RGB
vehicle movement or identifying stolen vehicles [22]. to grayscale simplifies the data, reducing the
 Image Processing Techniques: This section delves into computational complexity as the subsequent operations
the sophisticated image processing techniques that form need to process only one layer of pixel data instead of
the foundation of effective Automated Number Plate three [24].
Recognition systems, focusing on image acquisition,  Thresholding: This technique is used to create a binary
preprocessing methods, and advanced edge detection image from a grayscale image. It helps to differentiate
[22]. the foreground (number plate) from the background,
 Image Acquisition: Image acquisition is the initial and making the number plate more distinct [26].
critical step in the ANPR system, where digital cameras  Contrast Enhancement: Adjusting the contrast of the
are deployed to capture real-time images of vehicles on image can help in making the number plate more
the road. These cameras need to be high resolution to readable, especially in conditions where lighting does
ensure that the images captured are clear and detailed not favor clear visibility [27].
enough to identify the number plates accurately, even
under varying lighting and weather conditions. The  Advanced Edge Detection: Edge detection is a crucial
placement of these cameras is also strategic, ensuring technique in identifying the boundaries and structures
maximum coverage and optimal angles for capturing the within images, which is essential for locating number
number plates [23]. plates in the captured images. The chapter explores
advanced edge detection techniques.[18]
 Preprocessing Techniques: Once the images are
captured, they undergo several preprocessing steps to

Innovative Research Publication 31

 International Journal of Innovative Research in Computer Science and Technology (IJIRCST)

 Sobel Operator: A popular method used in edge precision, particularly in busy or complicated surroundings
detection that works by calculating the gradient of [20].
image intensity at each pixel within the image. It Advanced Image Processing: Using complex image
highlights regions of high spatial frequency that processing algorithms and high-resolution cameras to
correspond to edges [12]. manage a range of illumination scenarios and
 Contour Detection: Following edge detection, contour environmental obstacles [28].
detection is used to find a continuous line in the binary
image that outlines the number plate. This method helps V. OPTICAL CHARACTER RECOGNITION
in isolating the number plate from the rest of the image (OCR) FOR ANPR
[26]. This section explores the crucial role of Optical Character
 Region of Interest (ROI) Extraction: After detecting Recognition (OCR) in Automated Number Plate
contours, the specific area enclosing the number plate is Recognition systems, detailing the technologies involved,
extracted as the Region of Interest for further processing their integration within ANPR systems, and the practical
like character segmentation [1]. challenges encountered along with their solutions [5].
A. OCR Technologies
IV. LICENSE PLATE DETECTION WITH
FASTER R-CNN Optical Character Recognition (OCR) technologies are
designed to convert different types of documents, such as
This section focuses on the application of convolutional scanned paper documents, PDF files or images captured by
neural networks, particularly Faster R-CNN, to enhance the a digital camera into editable and searchable data. In the
accuracy and efficiency of license plate detection within context of ANPR, OCR technology is specifically used to
ANPR systems [11]. read and convert vehicle license plates into text. Key
technologies include:
A. Introduction to CNN and Faster R-CNN
Tesseract OCR: Widely used in ANPR systems due to its
Convolutional Neural Networks (CNNs) are a category of open-source nature and strong community support.
deep neural networks most commonly applied to analyzing Tesseract is capable of recognizing characters from images
visual imagery [12]. They are particularly powerful for with high accuracy, which is critical for the reliability of
tasks involving image recognition, classification, and object ANPR systems [16].
detection due to their ability to automatically detect Neural Network-based OCR: Advances in deep learning
important features without any human supervision. Faster have led to the development of OCR systems based on
R-CNN, an advanced version of CNN, stands out in the neural networks that provide improved accuracy over
field of object detection for its speed and efficiency [13]. It traditional methods, especially in challenging conditions
integrates a region proposal network (RPN) that shares full- such as poor lighting or atypical font styles[17].
image convolutional features with the detection network,
thus enabling nearly cost-free region proposals [14]. Faster B. Integration of OCR in ANPR Systems
R-CNN improves the speed of the process by connecting Integrating OCR technology into ANPR systems involves
the RPN directly to the last convolutional layer of a several steps:
standard CNN, making it a suitable choice for real-time Preprocessing: The image of the license plate is pre-
object detection tasks such as license plate recognition [16]. processed to enhance the quality and readability of the text.
This includes steps such as noise reduction, normalization,
B. Applying Faster R-CNN to ANPR
and binarization [21].
Faster R-CNN has been tailored specifically for the task of Character Segmentation: After preprocessing, OCR
license plate detection by training it on a diverse dataset of technology segments the image into individual characters.
vehicle images captured under different conditions. The Effective segmentation is crucial as it directly impacts the
process involves: OCR's ability to correctly recognize each character[22].
Training: The Faster R-CNN model is trained on a labeled Text Recognition: The segmented characters are then fed
dataset where the license plates are marked in various into the OCR engine, which interprets them as text. This
images. This training allows the model to learn the specific step is vital as the accuracy of the OCR affects the overall
characteristics of license plates, such as size, shape, and effectiveness of the ANPR system[23].
possible text configuration[13]. Data Verification and Storage: The recognized text is
Detection: During operation, when a new vehicle image is compared against vehicle registration databases for
captured, the trained Faster R-CNN model predicts the verification and stored for future reference or immediate
location of the license plate in the image. The model action, depending on the application[24].
generates region proposals where there is a high likelihood
of containing a license plate. These proposals are then C. Practical Challenges and Solutions
refined to accurately capture the boundaries of the license Despite the advancements in OCR technologies, several
plate[14]. practical challenges remain:
Integration: The detected license plates are then processed Variability in Plate Designs: License plates vary widely in
for character recognition and verification against databases terms of color, font style, and size, depending on the issuing
for vehicle identification and registration checks[15]. country or state. This variability can reduce OCR accuracy.
Solution: Implementing machine learning models that are
C. Recent Developments
trained on diverse datasets to improve the system's ability to
Multi-camera systems: These systems use several cameras generalize across different plate styles[25].
to record several viewpoints and improve detection

Innovative Research Publication 32

 International Journal of Innovative Research in Computer Science and Technology (IJIRCST)

Environmental Factors: Poor lighting, weather conditions, 𝑝x(x) = ∑ℎ−1 𝑓(x, j) ; 𝑝𝑦(𝑦) = ∑w−1 𝑓(i, 𝑦)
and fast-moving vehicles can impede the clarity of the
j=0 i=0
captured images. Solution: Utilizing robust preprocessing
techniques such as adaptive thresholding and employing The algorithm is currently iterating through each column of
more sophisticated imaging technologies like infrared the image in the step above. Each column's second pixel is
cameras[25]. the starting point, and it is compared to the threshold value.
Skewed or Partially Obscured Plates: Plates that are skewed The difference between the first and second pixel of the
or partially obscured pose recognition challenges. Solution: next column is the threshold value. Next, the difference
Deploying algorithms for image correction that adjust the between the second and third pixels is computed. If the
skew and enhance partially visible characters[26]. difference is more than the threshold, it is added to the total
Real-time Processing Requirements: ANPR systems are of the differences and continues in this manner until the end
often required to operate in real-time, which demands high of the column. At the conclusion, an array containing the
processing speeds. Solution: Optimizing OCR algorithms sum of the differences broken down by column is
for speed and implementing them on powerful hardware constructed. In terms of row-wise operation, the identical
platforms or via cloud-based architectures to ensure process is used to A vertical histogram can be found. An
minimal latency[27]. image's projection in both the vertical and horizontal
directions indicates a zone of interest in a certain direction
D. Image in grayscale = ( R+G+B)/3
[4].
Red component, R
G stands for green component. Accuracy: Under ideal circumstances, the ANPR systems
B is the blue component. examined in 2019 generally indicated accuracy rates of 85%
In [5], An image's edge is the location where its contrast or to 95%. When faced with difficult circumstances, such as
brightness changes abruptly or noticeably. The simplest dim illumination, unclean dishes, or obstacles, accuracy
basic tool for image identification is edge detection. may suffer [5].
Consequently, in order to identify every edge in the image, Efficiency: Real-time processing capabilities are the goal of
we must first reduce noise and sharpen the image. After many contemporary ANPR systems, with processing times
that, we mask it to identify the edges. A method for doing on high-performance hardware ranging from 100
this is known as the mask of Sobel Edge Detection. milliseconds to one second per frame [5].
Here, the mask used for sobel edge detection is
Garbage region filtering in an image: Following the above
Gx=sensing horizontal edges
procedure, a filter is applied to the image to eliminate the
Gy=identification of vertical edges
trash value region—unwanted parts. In this instance, the
The writer thought about the two masks for the input
rows and columns with lower histogram values are the
picture convolution. It is supplied to the horizontal and
undesirable regions [4]. A region's lower histogram value
vertical masks to convolve with the input image.
suggests that there aren't many differences between its
neighbouring pixels. Given that the license plate region is in
the foreground and has alpha numeric characters, its
histogram value will be very high due to the large
differences between its neighboring pixels.
Interest region Removal: Presently, the dynamic
thresholding technique is utilized, eliminating any areas
with lower histogram values and The region of interest is
the sole area that is remaining and has the greatest
E. Image Dilation histogram value. This is the license plate area.
Image Dilation is the process of sharpening the boundaries The average value of a histogram is the same as dynamic
and edges of an image to improve its quality. An image's thresholding. In this case, the average of the horizontal and
brightness is also increased. Additionally, it can eliminate vertical histograms is determined. The region of interest is
noise that might impede plate detection procedures. defined as the one with the highest histogram value, and it
Hence, it improves edge detection by sharpening the image. is retrieved from the data [11].
The image of a license plate must have the same levels of
brightness and contrast in order to be used for number plate
VI. LIMITATIONS
detection. For this reason, it's crucial to transform the
provided RGB image to greyscale. A. Variability in Plate Designs
Some crucial elements, such as brightness, finer details in It is challenging to develop an ANPR system that is
the object, sharper edges, etc., could be lost in the worldwide because different nations and areas have
converting process. A procedure called image dilation will different plate designs, formats, and typefaces[4].
make up for these losses.Detection of both vertical and
horizontal edges of a picture: A histogram is a graphical B. Environmental Elements
depiction of the distribution of greyscale values, both row- Weather (rain, snow, fog), shadows, and poor lighting
and column-wise, across adjacent pixels in an image. Both can all have an impact on an ANPR system's
horizontal and vertical histograms are used in this accuracy[5].
approach[13].
Let Px(x) be the horizontal projection of every pixel.
Let Py(y) be the vertical projection of every pixel.

Innovative Research Publication 33

 International Journal of Innovative Research in Computer Science and Technology (IJIRCST)

C. Distortion and Occlusion CNN," Measurement: Sensors, vol. 27, p. 100761, 2023.
Available from:
In addition to being partially obscured by debris,
https://doi.org/10.1016/j.measen.2023.100761
objects, or damage, plates can also be warped by motion [4] M. Mozumder, S. Biswas, L. Vijayakumari, R. Naresh, C. V.
blur or the camera's viewpoint[6]. Kumar, and G. Karthika, "A Hybrid Edge Algorithm for
D. Privacy Issues Vehicle License Plate Detection," in International
Conference on Intelligent Sustainable Systems, Singapore:
Growing surveillance gives rise to worries about privacy Springer Nature Singapore, pp. 209-219, Feb. 2023.
and possible data exploitation, which poses ethical and Available from: https://doi.org/10.1007/978-981-99-1726-
legal challenges[7]. 6_16
[5] M. Khinchi and C. Agarwal, "A Review on Automatic
E. Data Storage and Management Number Plate Recognition Technology and Methods," in
Managing substantial amounts of high-definition video Proc. International Conference on Intelligent Sustainable
data necessitates substantial processing and storage Systems (ICISS), pp. 363-366, 2019. Available from:
capacities [8]. https://doi.org/10.1109/ISS1.2019.8908014
[6] S. Sasi, S. Sharma, and A. N. Cheeran, "Automatic Car
F. Cost and Implementation Number Plate Recognition," in Proc. International
Especially for smaller governments or organizations, Conference on Innovations in Information, Embedded and
Communication Systems (ICIIECS), pp. 1-6, 2017, Available
the high price of implementing and maintaining cutting- from: https://doi.org/10.1109/ICIIECS.2017.8275893.
edge ANPR systems might be a barrier [5]. [7] A. Gautam, D. Rana, S. Aggarwal, S. Bhosle, and H.
G. Integration with Legacy Systems Sharma, "Deep Learning Approach to Automatically
Recognise License Number Plates," Multimedia Tools and
The efficacy of ANPR may be restricted by difficulties Applications, vol. 82, no. 20, pp. 31487-31504, 2023.
integrating it with current systems and infrastructure[2]. Available from: https://doi.org/10.1007/s11042-023-15020-
w
VII. CONCLUSION & FUTURE SCOPE [8] R. K. Prajapati, Y. Bhardwaj, R. K. Jain, and K. K. Hiran,
"A Review Paper on Automatic Number Plate Recognition
This paper has thoroughly explored the development and Using Machine Learning: An In-Depth Analysis of Machine
implementation of Automated Number Plate Recognition Learning Techniques in Automatic Number Plate
(ANPR) systems, highlighting their significance in Recognition: Opportunities and Limitations," in 2023
enhancing vehicle surveillance and public security[1]. International Conference on Computational Intelligence,
Utilizing Faster R-CNN, a cutting-edge deep learning Communication Technology and Networking (CICTN), pp.
527-532, Apr. 2023. Available from:
framework, ANPR systems have demonstrated a profound
https://doi.org/10.1109/CICTN57981.2023.10141318
capability in accurately detecting and recognizing vehicle [9] R. Mittal and A. Garg, "Text Extraction Using OCR: A
license plates under a variety of conditions[2]. Through Systematic Review," in Proc. Second International
detailed descriptions of the system’s architecture, from Conference on Inventive Research in Computing
image acquisition and pre-processing to the integration of Applications (ICIRCA), pp. 357-362, 2020, Available from:
Optical Character Recognition (OCR), this discussion https://doi.org/10.1109/ICIRCA48905.2020.9183326
underscores the technological sophistication and operational [10] F. Ali, H. Rathor, and W. Akram, "License Plate
effectiveness of modern ANPR systems[3]. Recognition System," in Proc. International Conference on
The future scope is that the automatic vehicle recognition Advance Computing and Innovative Technologies in
Engineering (ICACITE), pp. 1053-1055, 2021.
system plays a major role in detecting threats to defence
[11] N. P. Ap, T. Vigneshwaran, M. S. Arappradhan, and R.
also it can improve the security related to women as they Madhanraj, "Automatic Number Plate Detection in Vehicles
can easily detect the number plate before using a cab or sUsing Faster R-CNN," in Proc. International Conference
other services[6]. The system robustness can be increased if on System, Computation, Automation and Networking
a brighter and sharper camera is used withproper (ICSCAN), pp. 1-6, 2020. Available from:
installation[7]. https://doi.org/10.1109/ICSCAN49426.2020.9262400
[12] R. Laroca et al., "On the Cross-Dataset Generalization in
CONFLICTS OF INTEREST License Plate Recognition,"
https://doi.org/10.48550/arXiv.2201.00267.
The authors declare that they have no conflicts of interest. [13] L. Mufti and S. A. A. Shah, "Automatic Number Plate
Recognition: A Detailed Survey of Relevant Algorithms,"
Sensors, vol. 21, no. 9, p. 3028, 2021. Available from:
REFERENCES https://doi.org/10.3390/s21093028
[1] A. S. Tote, S. S. Pardeshi, and A. D. Patange, "Auto3.matic [14] V. Gnanaprakash, N. Kanthimathi, and N. Saranya,
Numb Plate Detection Using TensorFlow in Indian Scenario: "Automatic Number Plate Recognition Using Deep
An Optical Character Recognition Approach," Materials Learning," in IOP Conference Series: Materials Science and
Today: Proceedings, vol. 72, pp. 1073-1078, 2023. Engineering, vol. 1084, no. 1, 2021, IOP Publishing.
Available from: https://doi.org/10.1016/j.matpr.2022.09.165 Available from: https://doi.org/10.1088/1757-
[2] Y. Shambharkar, S. Salagrama, K. Sharma, O. Mishra, and 899X/1084/1/012027
D. Parashar, "An Automatic Framework for Number Plate [15] J. Tang et al., "Automatic Number Plate Recognition
Detection Using OCR and Deep Learning Approach," (ANPR) in Smart Cities: A Systematic Review on
International Journal of Advanced Computer Science and Technological Advancements and Application Cases,"
Applications, vol. 14, no. 4, 2023. Available from: Cities, vol. 129, 2022. Available from:
https://tinyurl.com/3zyppk95 https://doi.org/10.1016/j.cities.2022.103833
[3] M. A. Jawale, P. William, A. B. Pawar, and N. Marriwala, [16] J. M. Ravi Kumar, B. Sujatha, and N. Leelavathi,
"Implementation of Number Plate Detection System for "Automatic Vehicle Number Plate Recognition System
Vehicle Registration Using IoT and Recognition Using Using Machine Learning," in IOP Conference Series:

Innovative Research Publication 34

 International Journal of Innovative Research in Computer Science and Technology (IJIRCST)

Materials Science and Engineering, vol. 1074, no. 1, 2021,

IOP Publishing. Available from:
https://doi.org/10.1088/1757-899X/1074/1/012012
[17] J. Shashirangana, "Automated License Plate Recognition: A
Survey on Methods and Techniques," IEEE Access, vol. 9,
pp. 11203-11225, 2020. Available from:
https://doi.org/10.1109/ACCESS.2020.3047929
[18] A. O. Salau, T. K. Yesufu, and B. S. Ogundare, "Vehicle
Plate Number Localization Using a Modified GrabCut
Algorithm," Journal of King Saud University-Computer and
Information Sciences, vol. 33, no. 4, pp. 399-407, 2021.
Available from:https://doi.org/10.1016/j.jksuci.2019.01.011
[19] B. Mouli Chandra et al., "Recognition of Vehicle Number
Plate Using Matlab," Journal of University of Shanghai for
Science and Technology, vol. 23, no. 2, pp. 363-370, 2021.
Available from: https://jusst.org/wp-
content/uploads/2021/02/RECOGNITION-OF-VEHICLE-
NUMBER-PLATE-USING-MATLAB.pdf
[20] P. Kaur et al., "Automatic License Plate Recognition System
for Vehicles Using a CNN," Computers, Materials &
Continua, vol. 71, no. 1, 2022. Available from:
https://shorturl.at/JEtLF
[21] N.-A. Alam et al., "Intelligent System for Vehicles Number
Plate Detection and Recognition Using Convolutional
Neural Networks," Technologies, vol. 9, no. 1, 2021.
Available from:
https://doi.org/10.3390/technologies9010009
[22] S. M. Silva and C. R. Jung, "A Flexible Approach for
Automatic License Plate Recognition in Unconstrained
Scenarios," IEEE Transactions on Intelligent Transportation
Systems, 2021. https://doi.org/10.1109/TITS.2021.3055946
[23] C. Henry, S. Y. Ahn, and S.-W. Lee, "Multinational License
Plate Recognition Using Generalized Character Sequence
Detection," IEEE Access, vol. 8, 2020 Available from:
https://doi.org/10.1109/ACCESS.2020.2974973
[24] D. M. F. Izidio et al., "An Embedded Automatic License
Plate Recognition System Using Deep Learning," Design
Automation for Embedded Systems, 2020 Available from:
https://doi.org/10.1007/s10617-019-09230-5.
[25] V. Kukreja, D. Kumar, and A. Kaur, "GAN-Based Synthetic
Data Augmentation for Increased CNN Performance in
Vehicle Number Plate Recognition," in 2020 4th
International Conference on Electronics, Communication
and Aerospace Technology (ICECA), IEEE, 2020. Available
from: https://doi.org/10.1109/ICECA49313.2020.9297625
[26] A. Hamdi, Y. K. Chan, and V. C. Koo, "A New Image
Enhancement and Super Resolution Technique for License
Plate Recognition," IEEE Transactions on Intelligent
Transportation Systems, 2021. Available from:
https://www.cell.com/heliyon/fulltext/S2405-
8440(21)02444-0
[27] R. D. Castro-Zunti, J. Yépez, and S.-B. Ko, "License Plate
Segmentation and Recognition System Using Deep Learning
and OpenVINO," IET Intelligent Transport Systems, vol. 14,
no. 2, 2020. Available from: https://doi.org/10.1049/iet-
its.2019.0481
[28] C. Tu and S. Du, "A Hierarchical RCNN for Vehicle and
Vehicle License Plate Detection and Recognition,"
International Journal of Electrical and Computer
Engineering, vol. 12, no. 1, 2022. Available from:
https://doi.org/10.11591/ijece.v12i1.pp731-737

Innovative Research Publication 35