ET3491 – EMBEDDED SYSTEM AND IOT DESIGN

A MINI PROJECT REPORT

VECHILE NUMBER PLATE DETECTION

Submitted By:
THAMIZHARASAN B 212922106048
VIGNESH O 212922106051
THIRUVENGATAKRISHNAN M 212922106050
VIKRAM S 212922106053
SANJAY V 212922106036

In Partial Fulfilment for Award of

The Degree or Bachelor of Engineering in Electronics

and Communication Engineering at

St Joseph College of Engineering, Sriperumbudur

1
Anna University Chennai 602 117

BONAFIDE CERTIFICATE
Certified that this project report (VEHICLE NUMBER PLATE DETECTION) is the
Bonafide work of THAMIZHARASAN B (212922106048) VIGNESH O
(212922106051) THIRUVENGATAKRISHNAN M (212922106050) VIKRAM S
(212922106053) SANJAY V (212922106036) Carried out the project work under my
supervision

SIGNATURE: SIGNATURE:

MR.R. DINESH KUMAR M.E., (Ph.D.) MRS.D. ANGELINJENNIFER

HEAD OF THE DEPARTMENT ASSISTANT PROFESSOR

DEPARTMENT OF ECE DEPARTMENT OF ECE

ST. JOSEPH COLLEGE OF ST. JOSEPH COLLEGE OF

ENGINEERING ENGINEERING

CHENNAI-602 117 CHENNAI-602 117

Submitted for the semester Mini Project Viva Voice Examination held on _______

INTERNAL EXAMINER EXTERNAL EXAMINER

2
 ABSTRACT

In the last couple of decades, number of vehicles has been increased drastically.

Hence it has become very difficult to keep track of each and every vehicle for the

purpose of traffic management and the law enforcement. An automated, fast, accurate

and robust vehicle plate recognition system has become need for traffic control and law

enforcement of traffic regulations; and the solution is ANPR. Automatic Number Plate

Recognition (ANPR), is software used to recognize the number plates automatically by

performing sophisticated optical character recognition on images to read the number

plates of vehicles. In this work, methods such as number plate detection, optical

character recognition (OCR) were implemented. The number plate is detected using

YOLOv7 object detection model. This model is trained using more than 1500 images.

The proposed system can be mainly used to monitor road traffic activities such as the

identification of vehicle during traffic violations such as speed of vehicle and to detect

at the street traffic signals lane violation. And thereby can be traced every vehicle for

traffic rule violation and can provide the information to the concern authority to take

further effective action, so we can have smooth traffic flow and also, we can avoid

accidents occurring on the traffic junction. It is promising according to the performance

results that the proposed method will keep track of large number of vehicles and lead

to automation of vehicle management.

3
 TABLE OF CONTENTS

CHAPTER TITLE PAGE NO.

NO.

ABSTRACT 3

LIST OF ABBREVIATIONS 6

LIST OF FIGURES 7

1. INTRODUCTION

1.1 OVERVIEW 8
1.2 INTRODUCTION TO YOLO ALGORITHM 9

1.2.1 FEATURES 10

1.2.2 WORKING OF YOLO ALGORITHM 10

1.3 YOLOv7 12

1.3.1 YOLOv7 ARCHITECTURE 12

1.4 OPTICAL CHARACTER RECOGNITION (OCR) 16

1.4.1 WORKING OF OCR 17

1.4.2 USES AND ADVANTAGES OF OCR 18

1.4.3 PADDLEOCR 18

2. OVERVIEW OF THE PROJECT

2.1 METHODOLOGY 19

4
 2.2 DETECTION OF LICENSE PLATE USING YOLOv7 20

2.2.1 DATASET 20

2.2.2 ANNOTATING THE DATASET 20

2.2.3 TRAINING 21

2.2.4 OUTPUT OF TRAINED MODEL 21

2.3 PERFORMING OPTICAL CHARACTER RECOGNITION

2.3.1 PADDLEOCR ARCHITECTURE 22

2.3.2 CRNN ARCHITECTURE 23

2.3.4 IMPLEMENTATION 26

3. RESULTS AND DISCUSSIONS

3.1 EVALUATION OF THE TRAINED MODEL 28

3.2 ANPR OUTPUTS 33

4. CONCLUSION

4.1 DISCUSSION 34

4.2 FUTURE WORKS 35

REFERENCES 36

LIST OF ABBREVIATIONS

TERM ABBREVIATIONS

ANPR AUTOMATIC NUMBER PLATE RECOGNITION

5
YOLO YOU ONLY LOOK ONCE

OCR OPTICAL CHARACTER RECOGNITION

CNN CONVOLUTIONAL NEURAL NETWORKS

IOC INTERSECTION OVER UNION

E-ELAN EXTENDED EFFICIENT LAYER AGGREGATION

NETWORKS

NAS NETWORK ARCHITECTURE SEARCH

CRNN CONVOLUTIONAL AND RECURRENT NEURAL

NETWORKS

RNN RECURRENT NEURAL NETWORKS

CTC CONNECTIONIST TEMPORAL CLASSIFICATION mAP

MEAN AVERAGE PRECISION

LIST OF FIGURES

FIGURE TITLE PAGE NO.

NO.

1.1 PERFORMANCE OF YOLOv7 12

6
1.2 E-ELAN ARCHITECTURE 13

1.3 YOLOv7 COMPOUND SCALING 14

2.1 METHODOLOGY OF ANPR 19

2.2 DATASET 20

2.3 OUTPUT OF YOLOv7 22

2.4 PADDLEOCR ARCHITECTURE 22

2.5 CRNN ARCHITECTURE 23

2.6 FEATURE EXTRACTION 25

2.7 PADDLEOCR OUTPUT 27

3.1 CONFUSION MATRIX 29

3.2 PRECISION CURVE 30

3.3 RECALL CURVE 30

3.4 PRECISION-RECALL CURVE 31

3.5 F1 SCORE PLOT 31

3.6 RESULTS OF TRAINED MODEL 32

3.7 ANPR OUTPUTS 33

7
CHAPTER 1

INTRODUCTION

1.1 OVERVIEW

Traffic Management systems are installed on traffic signals to check for

vehicles breaking the traffic rules. In order to automate these processes and make
them more effective, a system is required to easily identify a vehicle. The important
question here is how to identify a particular vehicle, the obvious answer to this
question is by using the vehicle’s number plate as every vehicle has a unique
number through which it is easily differentiated from other vehicles.

Vehicles in each country have a unique license number, which is written on

its license plate. This number distinguishes one vehicle from the other, which is
useful especially when both are of same make and model. An automated system
can be implemented to identify the license plate of a vehicle and extract the
characters and numbers from the region containing a license plate.

The license plate number can be used to retrieve more information about the
vehicle and its owner, which can be used for further processing. Such an automated
system should be small in size, portable and be able to process data at sufficient
rate. Various number plate detection algorithms have been developed in past few
years.

Automatic Number Plate Recognition (ANPR) is the technology responsible

for reading the Number plates of a vehicle in an image or a video sequence using
optical character recognition. With the latest advancement in Deep Learning and
Computer Vision, these tasks can be done in a matter of milliseconds. The main

8
objective of the proposed design is to detect a license plate number from an image
which is captured from camera. An efficient algorithm is proposed to detect a
license plate under various conditions.

1.2 INTRODUCTION TO YOLO ALGORITHM

YOLO is an algorithm that uses neural networks to provide real-time object

detection. This algorithm is popular because of its speed and accuracy. It has been
used in various applications to detect traffic signals, people, parking meters, and
animals. Object detection in YOLO is done as a regression problem and provides
the class probabilities of the detected images.

YOLO algorithm employs convolutional neural networks (CNN) to detect

objects in real-time. As the name suggests, the algorithm requires only a single
forward propagation through a neural network to detect objects. This means that
prediction in the entire image is done in a single algorithm run. The CNN is used
to predict various class probabilities and bounding boxes simultaneously.

The YOLO algorithm consists of various variants. Some of the common

ones include tiny YOLO, YOLOv3, YOLOv4, YOLOv5, YOLOv7.

1.2.1 FEATURES

• Speed: This algorithm improves the speed of detection because it can predict
objects in real-time.

9
• High accuracy: YOLO is a predictive technique that provides accurate results
with minimal background errors.
• Learning capabilities: The algorithm has excellent learning capabilities that
enable it to learn the representations of objects and apply them in object
detection.

1.2.2 WORKING OF YOLO ALGORITHM

YOLO algorithm works using the following three techniques:

• Residual blocks
• Bounding box regression
• Intersection Over Union (IOU)

Residual blocks

First, the image is divided into various grids. Each grid has a dimension of
S x S. The following image shows how an input image is divided into grids. The
grid cells are of equal dimension. Every grid cell will detect objects that appear
within them. For example, if an object centre appears within a certain grid cell,
then this cell will be responsible for detecting it.

10
Bounding box regression

A bounding box is an outline that highlights an object in an image. Every

bounding box in the image consists of the following attributes

• Width (bw)
• Height (bh)
• Class (for example, person, car, traffic light, etc.)- This is represented by the
letter c.
• Bounding box centre (bx, by)

YOLO uses a single bounding box regression to predict the height, width,
centre, and class of objects. It represents the probability of an object appearing
in the bounding box.

Intersection over union (IOU)

Intersection over union (IOU) is a phenomenon in object detection that

describes how boxes overlap. YOLO uses IOU to provide an output box that
surrounds the objects perfectly.

Each grid cell is responsible for predicting the bounding boxes and their
confidence scores. The IOU is equal to 1 if the predicted bounding box is the same
as the real box. This mechanism eliminates bounding boxes that are not equal to
the real box.

1.3 YOLOv7:

11
• YOLOv7 is the fastest and most accurate real-time object detection model for
computer vision tasks. YOLOv7 is the basic model that is optimized for
ordinary GPU computing.

• YOLOv7 provides a greatly improved real-time object detection accuracy

without increasing the inference costs.

• YOLOv7 requires several times cheaper computing hardware than other deep
learning models. It can be trained much faster on small datasets without any
pre-trained weights.

Figure 1.1 PERFORMANCE OF YOLOv7

1.3.1 YOLOv7 ARCHITECTURE:

E-ELAN (Extended Efficient Layer Aggregation Network) in YOLOv7

The E-ELAN is the computational block in the YOLOv7 backbone. It has

been designed by analyzing the following factors that impact speed and
accuracy.

• Memory access cost

• I/O channel ratio
12
• Element wise operation
• Activations
• Gradient path

Figure 1.2 E-ELAN architecture

Compound Model Scaling in YOLOv7:

Different applications require different models. While some need highly

accurate models, some prioritize speed. Model scaling is performed to suit these
requirements and make it fit in various computing devices. While scaling a model
size, the following parameters are considered.

• Resolution (size of the input image)

• Width (number of channels)
• Depth (number of layers)

13
 • Stage (number of feature pyramids)

NAS (Network Architecture Search) is a commonly used model scaling method. It

is used by researchers to iterate through the parameters to find the best scaling
factors. However, methods like NAS do parameter-specific scaling. The scaling
factors are independent in this case.

Figure 1.3 YOLOv7 compound scaling

Trainable Bag of Freebies in YOLOv7

BoF or Bag of Freebies are methods that increase the performance of a

model without increasing the training cost. YOLOv7 has introduced the following
BoF methods.

Planned Re-parameterized Convolution

Re-parameterization is a technique used after training to improve the model.

It increases the training time but improves the inference results. There are two
types of re-parametrizations used to finalize models, Model level and Module level
ensemble.

Model level re-parametrization can be done in the following two ways.

14
• Using different training data but the same settings, train multiple models. Then
average their weights to obtain the final model.
• Take the average of the weights of models at different epochs.

Module level re-parameterization has gained a lot of traction in research. In this

method, the model training process is split into multiple modules. The outputs are
ensembled to obtain the final model.

Coarse for Auxiliary and Fine for Lead Loss

YOLO architecture comprises a backbone, a neck, and a head. The head

contains the predicted outputs. YOLOv7 does not limit itself to a single head. It
has multiple heads to do whatever it wants. Deep Supervision, a technique used
by DL models uses multiple heads. In YOLOv7, the head responsible for final
output is called the Lead Head. And the head used to assist training in the middle
layers is called the Auxiliary Head.

With the help of an assistant loss, the weights of the auxiliary heads are
updated. It allows for Deep Supervision and the model learns better. These
concepts are closely coupled with the Lead Head and the Label Assigner.

Label Assigner is a mechanism that considers the network prediction results

together with the ground truth and then assigns soft labels. It’s important to note
that the label assigner generates soft and coarse labels instead of generating hard
labels.

Lead Head Guided Label Assigner and Coarse-to-Fine Lead Head Guided Label
Assigner

15
The Lead Head Guided Label Assigner encapsulates the following three concepts.

• Lead Head
• Auxiliary Head
• Soft Label Assigner

The Lead Head in the YOLOv7 network predicts the final results. Soft labels
are generated based on these final results. The important part is that the loss is
calculated for both the lead head and the auxiliary head based on the same soft
labels that are generated.

1.4 OPTICAL CHARACTER RECOGNITION (OCR):

OCR (optical character recognition) is the use of technology to distinguish printed

or handwritten text characters inside digital images of physical documents, such
as a scanned paper document. The basic process of OCR involves examining the
text of a document and translating the characters into code that can be used for data
processing. OCR is sometimes also referred to as text recognition.

OCR systems are made up of a combination of hardware and software that

is used to convert physical documents into machine-readable text. In this stage the
number plate characters on the number plate image are converted into encoded
texts. An Artificial Neural Network (ANN) based OCR algorithm for ANPR
application is presented.

1.4.1 WORKING OF OCR:

The OCR engine or OCR software works by using the following steps:

16
Image acquisition

A scanner reads documents and converts them to binary data. The OCR
software analyzes the scanned image and classifies the light areas as background
and the dark areas as text.

Pre processing

The OCR software first cleans the image and removes errors to prepare it
for reading. These are some of its cleaning techniques:

• Tilting the scanned document slightly to fix alignment issues during

the scan.
• Removing any digital image spots or smoothing the edges of text
images.
• Cleaning up boxes and lines in the image.
• Script recognition for multi-language OCR technology
Text recognition

The two main types of OCR algorithms or software processes that an OCR
software uses for text recognition re called pattern matching and feature extraction.

Pattern matching

Pattern matching works by isolating a character image, called a glyph, and

comparing it with a similarly stored glyph. Pattern recognition works only if the
stored glyph has a similar font and scale to the input glyph. This method works
well with scanned images of documents that have been typed in a known font.

Feature extraction

Feature extraction breaks down or decomposes the glyphs into features

such as lines, closed loops, line direction, and line intersections. It then uses these

17
features to find the best match or the nearest neighbor among its various stored
glyphs.

1.4.2 USES AND ADVANTAGES OF OCR:

• Operational efficiency
• Artificial intelligence solutions
• Simple optical character recognition software
• Intelligent character recognition software
• Intelligent word recognition

The main advantages of OCR technology are the following:

• saves time
• decreases errors
• minimizes effort

1.4.3 PADDLEOCR

PaddleOCR offers users multilingual practical OCR tools that help the users to
apply and train different models in a few lines of code. PaddleOCR offers a lot of
models in its toolkit, including PP-OCR, a series of high-quality pretrained OCR,
the latest algorithms such as SRN, and popular OCR algorithms like CRNN.

PaddleOCR also provides different kinds of models whether it’s lightweight

(a model which takes less memory) or heavyweight (a model which takes large
memory) along with its pre-trained weights to use freely.

CHAPTER 2

OVERVIEW OF THE PROJECT

18
2.1 METHODOLOGY:

Figure 2.1 METHODOLOGY OF ANPR

1. Detection: Firstly, an image or a frame of the video sequence is passed to the

detection algorithm from a camera or an already stored file, which detects the
license plate and returns the bounding box location of that plate.
2. Recognition: The OCR is applied to the detected license plate for recognizing
the characters of the plate and returns the characters in the same order in text
format. The output can be stored in a database or can be plotted on the image
for visualization.

2.2 DETECTION OF LICENSE PLATE USING YOLOv7:

2.2.1 DATASET:

19
 Data is the core of any AI application and one of the first and most important
steps. For training the YOLOv7 number plate detector, dataset of vehicles will be
used. Our dataset consists of 1200 training images and 300 validation images and
100 validation images in the YOLO format.

Figure 2.2 DATASET

2.2.2 ANNOTATING THE DATASET:

Data Annotation (sometimes called “Data Labelling”) refers to the active

labelling of Machine Learning model training datasets. This often means adding
target labels but can also stand for adding feature values or metadata. In this project
we have used roboflow to annotate images.

To make the model learn, it needs to be trained on the dataset .

20
FACTORS IN TRAINING A YOLOv7 MODEL:

image = size of images on which model will train; the default value is 640.

batch-size = batch size used for custom dataset training.

epochs = number of training epochs to get the best model data

= custom config file path

weights = pretrained yolov7 weights

2.2.4 OUTPUT OF TRAINED MODEL:

21
 (a)

(b) Figure 2.3 OUTPUT OF YOLOv7

2.3 PERFORMING OPTICAL CHARACTER RECOGNITION:

2.3.1 PADDLEOCR ARCHITECTURE:

Figure 2.4 PADDLEOCR ARCHITECTURE

2.3.2 CRNN ARCHITECTURE:
22
 PaddleOCR are based on CRNN and provide easy-to-use pretrained models.
CRNN is a combination of both Convolutional and Recurrent neural networks.
Hence the name Convolutional Recurrent Neural Network (CRNN). This network
consists of three layers, CNNs followed by RNNs and then the
Transcription layer. CRNN uses CTC or Connectionist Temporal Classification
Loss which is responsible for the alignment of predicted sequences.

Figure 2.5 CRNN ARCHITECTURE

Feature Extraction

The first layer is the convolutional neural network (CNN) which consists of
Convolutional and max-pooling layers. These are responsible for extracting
features from the input images and producing feature maps as outputs. To feed

output to the next layer, feature maps are first converted into a sequence of feature
vectors. According to the original paper, “Each feature vector of a feature sequence
23
is generated from left to right on the feature maps by column. This means the i-th
feature vector is the concatenation of the i-th columns of all the maps.”

Due to the feature extraction, each column of the feature maps corresponds to a
rectangular region of the input image, that region is called a receptive field. Each
feature vector in the feature sequence is associated with the receptive field and can
be called an appearance descriptor for that region. The feature sequence is now
passed to the next layer of RNNs.

Sequence Labelling

This layer is the Recurrent Neural Network (RNN) which is built on top of the
Convolutional Neural Network. In CRNN, two Bi-directional LSTMs are used in
the architecture to address the vanishing gradient problem and to have a deeper
network. The recurrent layers predict the label for each feature vector or frame in
the feature sequence received from CNN layers. Mathematically, the layers predict
label y for each frame x in feature sequence x = x1,.., xt.

Transcription

This layer is responsible for translating the per-frame predictions into a final
sequence according to the highest probability. These predictions are used to
compute CTC or Connectionist Temporal Classification loss which makes the
model learn and decode the output.

CTC loss

The output received from the RNN layer is a tensor that contains the probability of
each label for each receptive field. when Connectionist Temporal Classification
(CTC) loss comes in. CTC loss is responsible for training the network as well as

24
the inference that is decoding the output tensor. CTC works on the following major
principles:

Figure 2.6 FEATURE EXTRACTION

• Text encoding: CTC solves the issue when a character takes more than one
time step. CTC solves this by merging all the repeating characters into one.
And, when that word ends it inserts a blank character “-”. This goes on for
further characters. For example, in fig -04, ‘S’ in ‘STATE’ has three time
steps. The network might predict those time steps as ‘SSS’. Now, the CTC
will merge those outputs and predict the output as ‘S’. For the word, a
possible encoding could be SSS-TT-A-TT-EEE, Hence the output ‘STATE’.

• Loss Calculation: For a model to learn, loss needs to be calculated and

backpropagated into the network. Here the loss is calculated by adding up
all the scores of possible alignments at each time step, that sum is the
probability of the output sequence. Finally, the loss is calculated by taking a
negative logarithm of the probability, which is then used for back-
propagation into the network.
• Decoding: At the time of inference, we need a clear and accurate output. For
this, CTC calculates the best possible sequence from the output tensor or
matrix by taking the characters with the highest probability per time step.
Then it involves decoding which is the removal of blanks “-” and repeated
characters.

2.3.4 Implementation:
25
In the above code snippet, we have initialized PP-OCRv3 and the required weights
will be downloaded automatically. This package by default provides all of the
models of the system which are detection, angle classification and recognition. It
provides several arguments to access only the required functionalities.

• lang: The language which we want to recognise is passed here. For example,
en for English, ch for Chinese, french for French, etc. The OCR can
recognise English and Chinese by default.
• rec_algorithm: Takes the recognition algorithm to be used as arguments. The
OCR uses CRNN as its default recognition algorithm. det_algorithm:
Takes the text detection algorithm to be used as arguments. The OCR uses a
DB text detector as its default detector.
• use_angle_cls: Specifies if angle classifier is to be used or not and takes bool
as the argument.

The OCR is now initialized and can be used in just one line of code.

This function also takes some arguments.

• img: This is the first parameter in the ocr function. In this, the image array
or the image path is passed to perform OCR.
• det: Takes bool as an argument and specifies whether to use a detector or
not. rec: Takes bool as argument and specifies whether to use a recognizer
or not.

26
 • cls: Takes bool as argument and specifies whether to use an angle classifier
or not.

Figure 2.7 PADDLEOCR OUTPUT

It detected all the fields like the boat number, date, ID number and more which
are the key information here even while the text was at an angle.

CHAPTER 3 RESULTS AND

DISCUSSION

3.1 EVALUATION OF THE TRAINED MODEL:

It is very important to judge the trained model’s performance on unseen data. It is

a good way to know whether the model is performing well or overfitting. For

27
object detection tasks, one such metric is mean average precision in short known
as mAP.

MEAN AVERAGE PRECISION (mAP):

Mean Average Precision(mAP) is a metric used to evaluate object detection models

such as Fast R-CNN, YOLO, Mask R-CNN, etc. The mean of average precision
(AP) values is calculated over recall values from 0 to 1. mAP formula is based on
the following sub metrics:

• Confusion Matrix,

• Intersection over Union (IoU),

• Recall,

• Precision

CONFUSION MATRIX:

To create a confusion matrix, we need four attributes:

True Positives (TP): The model predicted a label and matches correctly as per
ground truth.

True Negatives (TN): The model does not predict the label and is not a part of the
ground truth.

False Positives (FP): The model predicted a label, but it is not a part of the ground
truth (Type I Error).

False Negatives (FN): The model does not predict a label, but it is part of the
ground truth. (Type II Error).

28
 (a)

(b)

Figure 3.1 CONFUSION MATRIX

Precision:

Precision measures how well you can find true positives (TP) out of all positive
predictions. (TP+FP).

29
 Figure 3.2 PRECISION CURVE

Recall:

Recall measures how well you can find true positives (TP) out of all predictions
(TP+FN).

Figure 3.3 RECALL CURVE

Precision-Recall Curve:

Precision-Recall curve is obtained by plotting the model's precision and recall

values as a function of the model's confidence score threshold.

30
 Figure 3.4 PRECISION-RECALL CURVE

F1 SCORE:

The F-measure is the weighted harmonic mean of precision (P) and recall
(R) of a classifier, taking α=1 (F1 score). It means that both metrics have the same
importance.

Figure 3.5 F1 SCORE PLOT

RESULTS OF YOLOv7 MODEL:

The prediction results of the trained model are shown.

31
 Figure 3.6 RESULTS OF TRAINED MODEL

INFERENCE ON EVALUATION OF THE TRAINED MODEL:

YOLOv7 is a significant advance in terms of speed and accuracy, and it matches

or even outperforms RPN-based models. The model is fast and dependable, and it
can now be used for anything.

MEAN AVERAGE PRECISION = 0.778

IOU THERSHOLD = 0.5

F1 SCORE = 0.77

PRECISION = 1.00

RECALL = 0.93

3.2 ANPR OUTPUTS:

32
 Figure 3.7 ANPR OUTPUTS
CHAPTER 4 CONCLUSION

4.1 DISCUSSION:

We built an ALPR or ANPR system with a speed of 45 to 60 FPS. Here, we

focussed on a two-step process:
33
 i) License plate detector ii) Extraction and
OCR of license plate detector.
The license plate got trained with pretty good accuracy of 90%. The
PaddleOCR’s PP-OCR works flawlessly, it being lightweight as well and pretty
accurate which gives a good trade-off between accuracy and speed. PaddleOCR
provides various models like SRN, heavyweight PPOCR and many more, which
can be used or even trained from scratch to achieve desirable results.

But the ideal method for our ALPR is to use a tracker with it, which keeps
the best OCR result out of all. Various other trackers like OpenCV trackers,
CenterTrack, Tracktor etc. which tackle different advanced problems like
occlusion, Re-id etc.

Parking management at many places requires a lot of human interaction, but

due to ALPR it can be reduced to zero. The camera installed in the parking lots can
recognize the plates at the entrance and store them in a database or allow the
vehicles only if they are in the database. At the time of exit, the plate can be
recognized again and can be charged accordingly.

Law Enforcement: The ALPR can be used by police or law enforcement to

identify any traffic violation by a vehicle at a traffic light using only a single
camera. It can also be used to identify any stolen or non-registered vehicles in real-
time.

Intelligent traffic management: Automated vehicle number plate detection

can be used throughout cities to detect over speeding vehicles, vehicles that drive
rashly, or any accidental occurrence. ANPR provides solutions for measuring and
analyzing area-related traffic data of a certain area or an entire city.

4.2 FUTURE WORKS:

34
 This ANPR system works quite well however, there is still room for
improvement. This ANPR system speed can be increase with high resolution
camera. Which can be able to capture clear images of the vehicle. The OCR method
is sensitive to misalignment and to different sizes, so we have to create different
kind of templets for different Reginal transport office (RTO) specifications. The
statistical analysis can also be used to define the probability of detection and
recognition of the vehicle number plate. At present there are certain limits on
parameters like speed of the vehicle, script on the vehicle number plate, skew in
the image which can be removed by enhancing the algorithms further.

Today advances technology took Automatic Number Plate Recognition (ANPR)

systems from hard to set up, limited expensive, fixed based applications to simple
mobile ones in which “point to shoot” method can be used. This is possible because
of the creation of software which ran on cheaper PC based and also non specialist
hardware in which their no need to give pre- defined direction, angels, speed and
size in which the plate would be passing the camera field of view. Also, Smaller
cameras which can read license plates at high speed, along with smaller, more
durable processors that can fit in police vehicles, allowed law enforcement officers
to patrol daily with the benefit of license plate recognition in real time.

REFERENCES:

[1] C. Anagnostopoulos, T. Alexandropoulos, V. Loumos and E. Kayafas, “

Intelligent traffic management through MPEG-7 vehicle flow surveillance,”
in Proceedings of IEEE International Symposium on Modern Computing,
2006, pp. 202-207.

[2] N. Mani, and B. Srinivasan, “Application of artificial neural network model

for optical character recognition,” presented at the IEEE International
Conference on Systems, Man, and Cybernetics, 1997.

35
[3] R. P. van Heerden, E. C. Botha, Optimization of Vehicle License Plate
Segmentation and Symbol Recognition, Department of Electrical, Electronic
and Computer engineering, University of Pretoria, South Africa, 2010.

[4] F. M. Rodrguez, X. F. Hermida, New Advances in Automatic Reading of

V.L.P.s (Vehicle License Plates), Proceedings in SPC- 2000 (Signal
Processing and Communications), Marbella, September 2000.

[5] F.M. Rodrguez, M. G. Saburido, J. L. A. Castro, New Methods for Automatic

Reading of V.L.P.s (Vehicle License Plates), SPPRA- 2002 (Signal
Processing Pattern Recognition and Applications), June 2002.

[6] A. E. Savakis, Adaptive Document Image Thresholding Using foreground

and Background Clustering, International Conference on Image Processing,
October 1998.

[7] Ondrej Martinsky, Algorithmic and mathematical Principles of automatic

number plate Recognition systems, Brno University of technology, 2007.

[8] Sushruth Shastry, Gunasheela G, Thejus Dutt, Vinay D S and Sudhir Rao
Rupanagudi, i - A novel algorithm for Optical Character Recognition (OCR),
IEEE 2013.

[9] W. Badawy, ”Automatic License Plate Recognition (ALPR): A State of the

Art Review”, 2012

[10] Sukhpreet Singh Optical Character Recognition Techniques: A Survey,

Journal of Emerging Trends in Computing and Information Sciences, Vol. 4,
No. 6 June 2013.

36
[11] T. Naito, T. Tsukada, K. Yamada, K. Kozuka, and S. Yamamoto, Robust
licenseplate recognition method for passing vehicles under outside
environment, Trans. Veh. Technol., vol. 49, no. 6, pp. 23092319, Nov. 2000.

[12] M. Usman Akram, Zabeel Bashir, Anam Tariq and Shoab A Khan, Geometric
Feature Points Based Optical Character Recognition, IEEE Symp. Industrial
Elec. & App., Sept. 2013.

[13] Dr Savita Gael and Savita Dabas, Vehicle Registration Plate Recognition
System Using Template Matching, IEEE 97B-1 -4799-1 607-B/13, 2013.

[14] N. Saleem, H. Muazzam, H. Tahir, and U. Farooq, “Automatic license plate

recognition using extracted features,” presented at the 2016 4th International
Symposium on Computational and Business Intelligence (ISCBI), 2016, pp. 221–
225.

[15] Aniruddh Puranic,Deepak K. T.,Umadevi V “ Vehicle Number Plate

Recognition System: A Literature Review and Implementation using Template
Matching “ , International Journal of Computer Applications (0975 – 8887)
Volume 134 – No.1, January 2016.

[16] M. M. Shidore, S. P. Narote “ Number Plate Recognition for Indian Vehicles

“ , IJCSNS International Journal of Computer Science and Network Security,
VOL.11 No.2, Feb. 2011.

[17] Naito, T. Tsukada, T. Yamada, K. Kozuka, K. and Yamamoto, S., "Robust

recognition methods for inclined license plates under various illumination
conditions outdoors",Proceedings IEEE/IEEJ/JSAI International Conference
on Intelligent Transport Systems, pp. 697-702, 1999.

37
[18] M. R. Bai, V.V. Krishna, and J. Sreedevi, "A new morphological approach
for noise removal cum edge detection," IJCSI International Journal of
Computer Science Issues, Vol. 7, Issue 6, November 2010.

[19] Hontani, H., and Koga, T., "Character extraction method without prior
knowledge on size and information",Proceedings of the IEEE International
Vehicle Electronics Conference (IVEC'01), pp. 67-72, 2001.

[20] Norizam Sulaiman,“Development of Automatic Vehicle Plate Detection

System”,3rd International Conference on System Engineering and
Technology, pp.130-135, IEEE 2013.

38