0% found this document useful (0 votes)
54 views7 pages

Vehicle Number Plate Detection Under Different Environmental Conditions

Uploaded by

suresh merugu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
54 views7 pages

Vehicle Number Plate Detection Under Different Environmental Conditions

Uploaded by

suresh merugu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

International Journal for Multidisciplinary Research (IJFMR)

E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: [email protected]

Vehicle Number Plate Detection Under Different


Environmental Conditions
Anugu Hansika1, Thaviti Sony2, Sohaib Ibrahim3
1,2,3
Student, Department of Artificial Intelligence and Machine Learning, Sphoorthy Engineering College

Abstract
The main aim of the project is to detect the number plate of the vehicle and display the corresponding
vehicle number. The motive of this project stems from the need for robust and reliable number plate
detection systems that can perform effectively in real-world scenarios. Existing techniques frequently
have trouble dealing with different lighting situations, like dimly lit areas, which can result in erroneous
or missed detections. Furthermore, unfavourable angles of the number plates might make the detection
issues worse. The problem statement of this project is to develop a number plate detection system that is
capable of accurately locating number plates on vehicles, irrespective of the prevailing environmental
conditions. To achieve this, we employ a multi-stage approach that combines computer vision
techniques and image processing methodologies. Our proposed approach consists of three stages which
includes: pre-processing, feature extraction and text extraction

Keywords: Machine learning, computer vision, image processing, feature extraction.

1. Introduction
The project aims to develop a robust vehicle number plate detection system capable of accurate
operation under diverse environmental conditions. Advanced image processing techniques will be
employed to tackle challenges such as varying lighting, different angles, and vehicle speed. The system's
successful implementation holds the potential to improve traffic management, law enforcement, and
intelligent transportation systems. By enhancing number plate detection capabilities, the project aims to
contribute to safer roads and more efficient transportation systems, making it an asset for various
domains reliant on accurate and reliable vehicle identification. The task area of this project focuses on
the detection and localization of number plates on vehicles which involves the use of image processing
techniques and involves handling challenging environmental factors, including varying lighting
conditions, different vehicle speeds, and number plates at different angles. Employing advanced image
processing methods such as edge detection, adaptive thresholding, and contour analysis will enable the
system to accurately extract number plates, even under adverse conditions. To evaluate the above-
mentioned task, we used different modules which includes cv2, NumPy, Py tesseract and matplotlib
where cv2 provides functions to process and manipulate the images, NumPy provides support for
multidimensional arrays and mathematical functions to work on arrays, Py tesseract is used to read text
and Matplotlib is used for visualization of images.

2. Problem Statement
The current state of vehicle number plate detection systems reveals inherent challenges that hinder their

IJFMR240214540 Volume 6, Issue 2, March-April 2024 1


International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: [email protected]

effectiveness in real-world scenarios. Existing technologies encounter difficulties when confronted with
diverse environmental conditions, impacting their accuracy and reliability. The key problems that
motivate the proposed project are:
• Variability in Lighting Conditions:Current systems struggle to consistently perform under varying
lighting situations, leading to erroneous or missed detections, particularly in dimly lit or high-glare
environments.
• Challenges with Different Angles and Perspectives:Detecting number plates becomes problematic
when vehicles exhibit skewed or non-standard angles, making it difficult for existing systems to
accurately locate and identify plates.
• Speed-Related Issues:The speed at which vehicles move can introduce motion blur, affecting the
precision of number plate detection. Existing systems may face challenges in accurately identifying
plates under varying speeds.
• Font Style and Plate Variation: Non-standardized font styles on number plates and variations in
plate designs pose additional complexities. Existing systems may struggle to adapt to diverse font
styles and plate formats.
• Adaptability to Environmental Conditions:The need for a robust system capable of operating
seamlessly across different environmental conditions, including adverse weather, diverse lighting,
and varying vehicle speeds, is evident.

3. Literature Survey
In 1990, Lotufo et al. has worked on automated number plate recognition (ANPR). In 1994, Lee et al.
conducted yet another study on the identification of Korean licence plates in the past . In their research,
the region of interest is first identified using neural networks for color classification. The classification
process makes use of the pixels' Hue, Lightness, and Saturation (HLS) values. In 2018, Number plate
detection was carried out by Surekha, however they did not record any character identification. In their
research, they have used morphological processing, cunning edge detection, and neural networks. A
97% accuracy rate was recorded. They, however, did not consider many environmental conditions in
their approach. Computer vision is a major area for vehicle number plate detection. Computer vision is a
specialized branch of AI that is concerned with enabling computers to interpret and understand visual
information from the world, much like human vision. The process typically involves a series of
computer vision techniques and algorithms to achieve accurate and efficient number plate detection and
the process includes image acquisition, image preprocessing, number plate localization, character
segmentation, optical character recognition (OCR), and post-processing. These techniques enable
machines to automate the detection and recognition of number plates, making vehicle number plate
detection systems possible. By leveraging these techniques, systems can achieve accurate and reliable
detection, enabling a wide range of applications in traffic management, law enforcement, security and
automated systems. Enabling computers to see and comprehend visual data from the environment, akin
to human vision, is the focus of computer vision, a specialized area of artificial intelligence. Image
acquisition, image pre processing, number plate localization, character segmentation, optical character
recognition (OCR), and post-processing are some of the computer vision techniques and algorithms that
are typically used in the process to achieve accurate and efficient number plate detection. Vehicle
number plate detection systems are made possible by these techniques, which allow machines to
automatically detect and recognize number plates. These methods provide a broad range of applications

IJFMR240214540 Volume 6, Issue 2, March-April 2024 2


International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: [email protected]

in traffic management, law enforcement, security, and automated systems by offering accurate and
dependable detection for systems.

4. Proposed System
To develop an image processing system that can deal with cross angled, low illuminated, different font
number plates or blurry plates. The goal is to create number plate detection systems that can reliably
operate in various environmental conditions, contributing to traffic management, surveillance, law
enforcement, and other applications. Continued advancements will drive the adoption of these systems,
making our roads safer and transportation systems more efficient. our system can detect the plate and
extract the text from the input image. To develop an image processing system that can deal with cross
angled, low illuminated, different font number plates or blurry plates. The goal is to create number plate
detection systems that can reliably operate in various environmental conditions, contributing to traffic
management, surveillance, law enforcement, and other applications. Continued advancements will drive
the adoption of these systems, making our roads safer and transportation systems more efficient. our
system can detect the plate and exte 7 developed systems of three stages as shown in ract the text from
the input image.

“Figure 1”

4.1: Image Preprocessing


The input is the image of the vehicle. The collected image is prepared for the entire plate identification
procedure during the preprocessing stage. The preprocessing aims to reduce computational complexity
while ensuring the image retains essential information for accurate identification. Techniques like gray
scaling convert the colored image into a grayscale representation, reducing data processing costs.
Adaptive thresholding segments the image into foreground and background regions, considering varying
lighting conditions, enhancing consistency. Noise removal techniques further refine the image,
eliminating unwanted artifacts that could interfere with the identification process. By effectively
preprocessing the image through these techniques, subsequent stages benefit from reduced complexity,
leading to faster and more accurate vehicle number plate detection and recognition. The preprocessing
stage is an indispensable part of automatic license plate recognition systems, as it lays the foundation for
reliable and efficient identification of number plates in various environmental conditions.

4.2: Feature Extraction


Feature extraction is a fundamental technique in computer vision and pattern recognition, aiming to
capture meaningful information from raw data, such as images, to reveal underlying patterns or
characteristics. In the context of pre-processed images, relevant features are extracted using methods like
edge detection and morphological operations. Edge detection identifies the boundaries of objects in the
image, representing prominent transitions in intensity, which can be crucial for recognizing object
shapes. Morphological operations manipulate image structures to fill gaps, remove noise, and highlight

IJFMR240214540 Volume 6, Issue 2, March-April 2024 3


International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: [email protected]

regions of interest. By extracting these relevant features, the processed data can be represented more
effectively, aiding subsequent analysis and tasks, such as object recognition, classification, or detection.
Feature extraction is a vital step in computer vision systems, as it enables the transformation of complex
data into meaningful representations that facilitate advanced pattern recognition and understanding of
visual information.

4.3: Text Extraction


Text extraction in vehicle number plate detection is facilitated by PyTesseract, a popular Python library
that interfaces with the Tesseract OCR engine. Tesseract OCR, developed by Google, is widely used for
accurate text extraction from images. In the context of vehicle number plates, PyTesseract can be
applied to recognize the alphanumeric characters present in the localized number plate region. By
extracting the text from the number plate, valuable information for vehicle identification and other
applications can be obtained. The seamless integration of PyTesseract with Tesseract OCR makes it a
valuable tool in the field of computer vision, enabling efficient and reliable text extraction from images.

5. Methodology

“Figure 2”

“Figure 3”

IJFMR240214540 Volume 6, Issue 2, March-April 2024 4


International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: [email protected]

5.1: Read Image


The imread function from OpenCV is essential for reading image files, enabling analysis and processing
of visual data. This step is vital to gain insights from the images and perform various tasks like object
detection, recognition, and more, contributing to computer vision and AI applications.

5.2: Gray Conversion


Gray scaling, also known as gray conversion, is an essential image preprocessing technique widely
employed in various computer vision applications, particularly in vehicle number plate detection. The
process involves converting a colored image into a grayscale image, where each pixel is represented by a
single intensity value instead of separate red, green, and blue (RGB) channels. The benefits of gray
scaling are multifaceted. By reducing the image's complexity, it helps improve the accuracy of
recognition algorithms, making them more efficient in identifying and processing visual data.
Additionally, gray scaling proves advantageous in mitigating the impact of inconsistent or varying
lighting conditions, which are common challenges in number plate detection tasks. Such lighting
variations can distort colors, making color-based recognition less reliable. By simplifying the input data,
gray scaling provides a consistent and standardized representation of images, facilitating subsequent
analysis, and enabling more accurate and reliable results in vehicle number plate detection and other
computer vision tasks. Fig 4 describes gray conversion .

5.3: Adaptive Thresholding


Adaptive thresholding is a vital image processing technique used in computer vision to binarize images,
separating foreground objects from the background. This method is particularly beneficial when dealing
with varying lighting conditions or uneven illumination. By converting the image into a binary format,
adaptive thresholding allows the identification of regions of interest (ROIs) essential for further analysis,
such as object recognition or tracking. The ROIs encompass the foreground objects, while the
background objects, including noise or irrelevant elements, are excluded. This distinction aids in
focusing on relevant information and improves the accuracy of subsequent computer vision tasks. Fig 5
visually demonstrates the process of adaptive thresholding, showcasing its significance in image
segmentation and its valuable role in various computer vision applications.

5.4: Morphological Operations


Morphological operations are commonly used in vehicle number plate detection to manipulate the shape
and structure of binary images. These operations help in removing noise, filling gaps, and enhancing the
quality of the detected number plate regions. Two commonly used morphological operations are dilation
and erosion. Dilation expands the boundaries of the regions in the binary image, while erosion shrinks
them. These operations can be adjusted by changing the size and shape of the kernel as well as the
number of iterations.

5.5: Contour Identification


The find Contours function identifies the contours in the processed image. Contours represent the
boundaries of objects in the image. Contours represent the boundaries of objects in an image, and by
identifying and analysing these contours, we can determine the shape and location of the number plates.
Once the relevant contours (regions of interest) are retained based on the area thresholding, the Hough

IJFMR240214540 Volume 6, Issue 2, March-April 2024 5


International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: [email protected]

Transform is applied to 14 further refine the identification of the number plate regions. The Hough
Transform is a technique used to detect straight lines or shapes in an image, such as the edges of the
number plate. By applying the Hough Transform, the number plate's potential boundary lines can be
identified, helping to confirm and precisely locate the number plate region. The combination of
threshold area filtering and the Hough Transform assists in accurately and efficiently detecting and
extracting the number plates from the processed image.

5.6: Text Extraction


It involves extracting the alphanumeric characters from the detected number plate region. One popular
approach for text extraction is to utilize Optical Character Recognition (OCR) techniques.

6. Conclusion
This study focuses on Vehicle number plate detection and recognition. The main contributions of this
study include considering difficult scenarios including change in lighting conditions, blurry, noisy
pictures, skewed and non-standard and partially worn-out number plates. The preprocessing of this study
includes image processing techniques including gray scaling, adaptive thresholding. Next, for
segmentation of the plate contours are applied. Finally, by iterating through the contours we identify our
region of interest to extract the vehicle number. Future technological and Artificial Intelligence
developments are projected to increase the efficiency of vehicle number plate recognition in a variety of
environmental settings. The following upcoming breakthroughs might strengthen the relationship
between AI and number plate detection: enhanced imaging and sensors, AI based image processing,
contextual understanding.

7. Acknowledgement
We express our deep sense of gratitude to our project Co-Ordinator Mr.Mohammed Afzal for his
constant and continuous support throughout the project. We sincerely thank our Project Guide
Mrs.L.Swathi, Asst. Professor, Department of Computer Science & Engineering (AI&ML), Sphoorthy
Engineering College, for her inspiring guidance, consistent encouragement, constructive criticism and
helpful suggestions during the entire course of my project work. We express our sincere thanks to Mrs.
S. Indhumathi, Assistant Professor & Head of the Department, Department of Computer Science &
Engineering (AI&ML), Sphoorthy Engineering College, Nadergul (V), Balapur (M), Ranga Reddy (D)
for her encouragement which helped us to complete the Project work. We deem it a great privilege to
express our profound gratitude and sincere thanks to Mr. S. Chalama Reddy, Chairman, Mr. S. Jagan
Mohan Reddy, Secretary, Dr. V. Venkata Krishna, Principal, Sphoorthy Engineering College, Nadergul
(V), Balapur (M),Ranga Reddy (D), for their moral support and help in the completion of the project
work. We express our heartfelt thanks to Professors, Associate Professors, Assistant Professor and other
professional non-teaching staff of Department of Computer Science and Engineering (AI&ML),
Sphoorthy Engineering College, Nadergul (V), Balapur (M), Ranga Reddy (D) for providing the
necessary information pertaining to our project work.

8. References
1. Ching-Tang Hsieh, Yu-Shan Juan and Kuo-Ming Hung. (2005) “Multiple License Plate Detection
for Complex Background,” 19th International Conference on Advanced Information Networking an-

IJFMR240214540 Volume 6, Issue 2, March-April 2024 6


International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: [email protected]

d Applications (AINA'05), Taipei.


2. P Surekha, Pavan Gurudath, R Prithvi and VG Ritesh Ananth. (2018) “Automatic license plate
recognition using image processing and neural networks,” ICTACT Journal on Image and Video
Processing (IJIVP), 8 (4): 1786-1792.
3. Ayodeji Olalekan Salaua, Thomas Kokumo Yesufua and Babatunde Sunday Ogundare, (2019)
“Vehicle plate number localization using a modified GrabCut algorithm,” Journal of King Saud
University - Computer and Information Sciences, In Press
4. P. William, A. Shrivastava, P.S. Chauhan, M. Raja, S.B. Ojha, K. Kumar, Natural Language
processing implementation for sentiment analysis on tweets, in: N. Marriwala, C. Tripathi, S. Jain,
D. Kumar (Eds.), Mobile Radio Communications and 5G Networks, Lecture Notes in Networks and
Systems, vol. 588, Springer, Singapore, 2023, https://doi.org/10.1007/978-981-19-7982-8_26.
5. S. Mishra, S. Choubey, A. Choubey, N. Yogeesh, J. Durga Prasad Rao, P. William, Data extraction
approach using natural language processing for sentiment analysis, in: International Conference on
Automation vol. 2022, Computing and Renewable Systems (ICACRS), Pudukkottai, India, 2022,
pp. 970–972, https://doi.org/ 10.1109/ICACRS55517.2022.10029216.
6. Deepak Narayan Paithankar, Abhijeet Rajendra Pabale, P. William Rushikesh Vilas Kolhe,
Prashant Madhukar Yawalkar, Framework for implementing air quality monitoring system using
LPWA-based IoT technique, Measurement: Sensors 100709 (2023),
https://doi.org/10.1016/j.measen.2023.100709. ISSN 2665-9174.
7. https://www.ijert.org/research/a-review-paper-on-vehicle-number-plate-recognition-
IJERTV8IS040246.pdf
8. https://www.sciencedirect.com/science/article/pii/S2665917423000971
https://www.scopus.com/record/display.uri?eid=2-s2.0-
84928103322&origin=inward&txGid=3e709bcd8d802288c16764033ece02ae

IJFMR240214540 Volume 6, Issue 2, March-April 2024 7

You might also like