SU+ @ Strathmore 

University Library 
 

  
 

Electronic Theses and Dissertations 

 

 

This work is availed for free and open access by Strathmore University Library.  

It has been accepted for digital distribution by an authorized administrator of SU+ @Strathmore University. 

For more information, please contact library@strathmore.edu 

 

 
2021 

 

A Duplicate number plate detection system 

using fixed location cameras. 

 
 
Kibunja, Kelvin Peter 
School of Computing and Engineering Science  
Strathmore University 
 
 
 
 

 
 
 
 
 
 
 
 
 

 

Recommended Citation 

Kibunja, K. P. (2021). A Duplicate number plate detection system using fixed location cameras  [Thesis, 

Strathmore University]. http://hdl.handle.net/11071/12755  

 
 

 
Follow this and additional works at: http://hdl.handle.net/11071/12755 

https://su-plus.strathmore.edu/
https://su-plus.strathmore.edu/
http://hdl.handle.net/11071/2474
mailto:library@strathmore.edu
http://hdl.handle.net/11071/12755
http://hdl.handle.net/11071/12755


A Duplicate Number Plate Detection System Using Fixed Location 

Cameras 

 

 

 

 

By 

 Kelvin Peter Kibunja 

114814 

 

 

 

A Thesis Submitted to the School of Computing and Engineering Science in partial fulfillment of 

the requirements for the award of Master of Science in Computing and Information Systems 

Degree. 

 

Master of Science in Computing and Information Systems 

 Strathmore University 

September 2021



 

 

ii 

 

Declaration 
I declare that this work has not been previously submitted and approved for the award of a degree 

by this or any other University. To the best of my knowledge and belief, the thesis contains no 

material previously published or written by another person except where due reference is made in 

the thesis itself. 

© No part of this thesis may be reproduced without the permission of the author and Strathmore 

University. 

Njeri, KelvinPeter Kibunja 

 

September 2021 

Approval 

The thesis of Njeri, KelvinPeter Kibunja was reviewed and approved by the following: 

Dr. Vitalis Ozianyi, 

Senior Lecturer, Scholl of Computing and Engineering Sciences, 

Strathmore University 

 

Dr. Julius Butime, 

Dean, School of Computing and Engineering Sciences, 

Strathmore University 

 

Dr. Bernard Shibwabo, 

Director of Graduate Studies, 

Strathmore University 

 

 



 

 

iii 

 

Abstract 
 

Vehicle registration is a legal necessity to all, any vehicle purchased must be registered and the 

purpose of this registration is to assist in vehicle identification. The vehicle identifier is usually a 

metal or plastic plate patched on the front or rear part of a vehicle and contains unique 

alphanumeric characters. If vehicles on the road do not have any unique identifying tags or have 

the wrong tags, people will have, an opportunity to be mischievous as no one will be able to 

identify or account for any vehicle. Hence, through vehicle registration one can be able to identify 

and prove ownership of the vehicle. In order to hide the identity vehicles are disguised by cloning 

the number plates which aids in providing inaccurate ownership details. Without the correct 

ownership details, it is difficult to trace any vehicle involved in an illegal activity be it theft, 

terrorism, to mention but a few. 

In Kenya for people to discover that a vehicle has duplicate number late, the vehicle must have 

attracted the police attention, and this could be through being involved in an illegal activity. This 

system of number plate identification is usually a posterior to a crime, and it entails conducting a 

search with the Kenya National Transport Safety Authority (NTSA). In a country like Britain, the 

system in place to identify number plates is a deterrent one, whereby fixed location cameras 

capture images of vehicle’s number plates then compare with what is in their regional registration 

database. In Kenya this system would work perfectly well but in cases where we have duplicates 

it will be an issue since it will be difficult to flag out one. 

To curb the problem, the researcher intends to use fixed location cameras to identify duplicate 

number plates. From the image, the system extracts the vehicle number plate, colour and model. 

Time and location of the image capture is recorded, and this will assist in comparison of the details 

captured. The researcher uses image processing and artificial neural network in identifying the 

colour and model of the vehicle. Prototyping methodology and analysis approach is used in 

developing the model as it allows development of independent modules on large scale applications. 

 



 

 

iv 

 

Table of Contents 
Declaration ...................................................................................................................................... ii 

Abstract .......................................................................................................................................... iii 

List of Tables ................................................................................................................................ vii 

List of Figures .............................................................................................................................. viii 

List of Abbreviations ..................................................................................................................... ix 

Chapter 1: Introduction ................................................................................................................... 1 

1.1 Background ...................................................................................................................... 1 

1.2 Problem Statement ................................................................................................................ 3 

1.3 Aim ........................................................................................................................................ 4 

1.4 Specific Objectives ................................................................................................................ 4 

1.5 Research Questions ............................................................................................................... 4 

1.6 Justification ........................................................................................................................... 4 

1.7 Scope and Limitation ............................................................................................................ 5 

Chapter 2: Literature Review .......................................................................................................... 6 

2.1 Introduction ........................................................................................................................... 6 

2.2 Vehicle Identification ............................................................................................................ 6 

2.3 Number Plate Identification Systems .................................................................................... 8 

2.3.1 Security Marking Technologies...................................................................................... 8 

2.3.2 Advanced Electronic Signature ...................................................................................... 9 

2.3.3 2D Barcoding ................................................................................................................ 10 

2.3.4 Radio Frequency Identification (RFID) ....................................................................... 11 

2.3.5 Automatic Number plate recognition. .......................................................................... 13 

2.4 Machine Learning to Identify the Number Plates ............................................................... 16 

2.4.1 Image Acquisition and Pre-process .............................................................................. 16 

2.4.2 Number Plate Detection and Extraction ....................................................................... 17 

2.4.3 Character Extraction and Segmentation ....................................................................... 18 

2.4.4 Vehicle Colour Recognition ......................................................................................... 18 

2.5 Challenges Associated with Current Automatic Number Plate Recognition Systems ....... 20 

2.6 Conceptual Model............................................................................................................ 21 

Chapter 3: Research Methodology................................................................................................ 23 



 

 

v 

 

3.1 Introduction ......................................................................................................................... 23 

3.2 Research Design .................................................................................................................. 23 

3.3 System Development Methodology .................................................................................... 23 

3.3.1 Prototyping Methodology ............................................................................................. 24 

3.4 Data Collection Instruments ................................................................................................ 27 

3.5 Database Development Tools.............................................................................................. 27 

3.6 Programming Tools ............................................................................................................. 27 

3.7 System Analysis .................................................................................................................. 28 

3.8 Research Quality ................................................................................................................. 28 

3.9 Ethical Considerations......................................................................................................... 29 

Chapter 4: System Design and Architecture ................................................................................. 30 

4.1 Introduction ......................................................................................................................... 30 

4.2 Requirement Analysis ......................................................................................................... 30 

4.2.1 Functional Requirements .............................................................................................. 30 

4.2.2 Non-functional Requirements ....................................................................................... 30 

4.3 System Design and Architecture ......................................................................................... 31 

4.3.1 Data Input ..................................................................................................................... 32 

4.3.2 Data Processing ............................................................................................................ 33 

4.3.3 Classification Output .................................................................................................... 33 

4.3.4 Process Design .............................................................................................................. 33 

4.4 System Design ..................................................................................................................... 33 

4.5 Data Flow Diagram ............................................................................................................. 33 

4.6 Use-Case Diagram............................................................................................................... 35 

4.6.1 Detailed description of Use-Case Diagram .................................................................. 36 

4.7 System Sequence Diagram .................................................................................................. 39 

Chapter 5: System Implementation and Testing ........................................................................... 41 

5.1 Overview ............................................................................................................................. 41 

5.2 Description of The Testing Environment ............................................................................ 41 

5.2.1 Hardware Specifications ............................................................................................... 41 

5.2.2 Software Specifications ................................................................................................ 42 

5.3 Prototype Development Environment ................................................................................. 42 



 

 

vi 

 

5.4 Model Components ............................................................................................................. 43 

5.4.1 Image Input Components .............................................................................................. 43 

5.4.2 System Components ..................................................................................................... 47 

5.5 System Testing .................................................................................................................... 48 

5.5.1 Camera Application ...................................................................................................... 48 

5.5.2 Storing Captured Information ....................................................................................... 51 

5.5.2 Flagging out of Duplicate Number Plates .................................................................... 53 

5.6 Performance Measure .......................................................................................................... 54 

5.6.1 Functional Requirements .............................................................................................. 54 

5.6.2 Usability Requirements ................................................................................................ 55 

5.6.3 Reliability Requirements .............................................................................................. 56 

5.6.4 Supportability Requirements ........................................................................................ 56 

Chapter 6: Discussion, Conclusions and Recommendations ........................................................ 57 

6.1 Introduction ......................................................................................................................... 57 

6.2 Evaluation of The Systems Used in Vehicle Identification ................................................ 57 

6.3 Analysis of Number Plate Identification ............................................................................. 57 

6.4 Challenges Associated with The Use of Image Recognition .............................................. 57 

6.5 Develop a System to Flag out Duplicate Number Plates Using Fixed Location Camera ... 58 

6.6 Conclusions ......................................................................................................................... 58 

6.7 Recommendations ............................................................................................................... 58 

6.8 Suggestions for Future Works ............................................................................................. 59 

References ..................................................................................................................................... 60 

Appendices .................................................................................................................................... 67 

Appendix A: Originality Report ................................................................................................ 67 

 

  



 

 

vii 

 

List of Tables 
Table 2.1: Numbers plate detection Methods ............................................................................... 20 

Table 4.1: Use Case Diagram. ...................................................................................................... 37 

Table 4.2: Image Analysis ............................................................................................................ 37 

Table 4.3: Vehicle Details Classification...................................................................................... 38 

Table 5.1: Hardware Specification. .............................................................................................. 42 

Table 5.2: Functional Requirements. ............................................................................................ 54 

Table 5.3: Usability Requirements. .............................................................................................. 55 

Table 5.4: Reliability Requirements. ............................................................................................ 56 

Table 5.5: Supportability Requirements. ...................................................................................... 56 

 

 

 

 

 

 

 

 

 

 

 

 

  



 

 

viii 

 

List of Figures  
Figure 1.1: Car Cloning Report (Wheatstone, 2019). ..................................................................... 5 

Figure 2.1: Digital Signature Process Flow (Hoffman et al., 2013). ............................................ 10 

Figure 2.2: 2D Barcoding (White & Gardiner, 2007). .................................................................. 11 

Figure 2.3: Components of an RFID System (Jia et al., 2012). .................................................... 12 

Figure 2.4:Reader and Tag Message Communication (Larionov et al., 2017). ............................ 13 

Figure 2.5: Image Capture from a Camera (Dias et al., 2019)...................................................... 14 

Figure 2.6: Binary Images (Kashyap et al., 2018). ....................................................................... 15 

Figure 2.7: Steps for Numbers Plate Identification From an Image (Kurdi & Ahmad, 2017). .... 16 

Figure 2.8: Conceptual Model. ..................................................................................................... 22 

Figure 3.1:Prototyping Development Model (Madhav mohan, 2019). ........................................ 25 

Figure 4.1: System Architecture. .................................................................................................. 32 

Figure 4.2: Data Flow Diagram. ................................................................................................... 34 

Figure 4.3: Level 0 Diagram. ........................................................................................................ 35 

Figure 4.4: Use Case Diagram. ..................................................................................................... 36 

Figure 4.5: Sequence Diagram. ..................................................................................................... 40 

Figure 5.1: Camera Home page. ................................................................................................... 44 

Figure 5.2: Upload Page. .............................................................................................................. 45 

Figure 5.3: Capture Page............................................................................................................... 46 

Figure 5.4: No Caption page. ........................................................................................................ 47 

Figure 5.5: System Login Page. .................................................................................................... 48 

Figure 5.6: Capture Page............................................................................................................... 49 

Figure 5.7: Upload Page. .............................................................................................................. 50 

Figure 5.8: Error Page. .................................................................................................................. 51 

Figure 5.9: Home page. ................................................................................................................. 52 

Figure 5.10: Data base. ................................................................................................................. 53 

Figure 5.11: Duplicate Number Plates. ......................................................................................... 54 

 

file:///C:/Users/kkibunja/Desktop/Thesis%202021/Final%20Document/A%20Duplicate%20Number%20Plate%20Detection%20System%20Using%20Fixed%20Location%20Cameras.docx%23_Toc71293564


 

 

ix 

 

List of Abbreviations 
ANN - Artificial Neural Networks 

ANPR - Automatic Number Plate Recognition  

CNN - Convolution Neural Network  

IDF - Import Declaration Form  

ITS - Intelligent Transport Systems  

NTSA - National Transport and Safety Authority  

PSV - Public Service Vehicles  

RFID - Radio Frequency Identification Technology  

SVM - Support vector machines  

 

 

 

 

 

 

 

 

 

 

 



 

 

x 

 

Acknowledgements 
 

First and foremost, I thank the Almighty God for seeing me through thick and thin in this life and 

for his grace. 

I have been particularly fortunate to have the superb guidance and academic mentorship of Dr. 

Vitalis Ozianyi even though he was handling several projects at the time, at no time did he falter 

in offering his wisdom, wit, and research expertise until the completion of this project. Thank you 

very much Sir I will remain indebted to you. My sincere gratitude to the members of School of 

Computing and Engineering sciences Dr Vincent Omwenga Dr Joseph Orero and Dr Bernard 

Shibwabo. 

To my adoring parents, Peter, and Anne Ekoropus, I extend my utmost appreciation and gratitude 

for giving me the gift of education and the life lessons that brought out the best in me. May the 

almighty God bless you and help you to live long enough to see the good work of your hands. 

To my siblings Ian, Brenda, Catherine Lavender, and Louis, I appreciate your love and assistance 

always.  

Special thanks to my friends; Tiberius, Kevin, Titus, Charles and Thuo for walking with me. 

Lastly, I would like to thank my best friend most sincerely, Nancy. 

 

 

 

 

 

 

 

 

 



 

 

xi 

 

Dedication 
This project is dedicated to my family members and friends for their understanding of the time 

taken during this study.  

 

 

 



 

 

1 

 

Chapter 1: Introduction 

1.1 Background  

Vehicle registration is a lawful act done to give identity to vehicles. This identity can be used to 

trace the details of the vehicles when need be and can also be used to identify vehicles that are 

owned by an individual, the state or any other organisation. The vehicle is given a unique identity 

as part of the process of registering vehicles (Driver & Vehicle Licensing Agency, 2014). The 

process of vehicle registration has been changing over the years because of the details being 

captured and the methods used in capturing the details. There is an authority that is usually 

mandated to handle registration. In Kenya registration, licensing of motor vehicles and drivers is 

handled by the National Transport and Safety Authority (NTSA) (David, 2017). 

After registration of the vehicle one is issued with a number plate depending on the ownership of 

the vehicle: Government of Kenya, parastatals, diplomats, local authorities and civilians. The 

process of registration should be handled by the vehicle dealers before civilians take ownership of 

the vehicle. The traffic act CAP 403 states that no motor vehicle imported for home use shall be 

used on a road unless it is registered. Thus, ownership of a car does not licence any one to use the 

car until registration is done hence one cannot use the vehicle on the Kenyan roads. It is illegal to 

have unregistered vehicle on the Road. 

For anyone to register a vehicle, one requires to provide well documented duty and VAT receipts, 

import entry form, foreign registration book which must be in English, Ainuays bill, bill of lading, 

Import Declaration Form (IDF), duplicate valid insurance certificate, Vehicle Inspection Report 

for Commercial and Public Service Vehicles (PSV) photocopies of the owner’s national identity 

card or business registration certificate where applicable, PIN card and the completed application 

form (Form A) (Procedures for Motor Vehicle - KRA, n.d.). After the registrar has gone through 

the submitted documents and they have ascertained that the documents and information provided 

certifies their requirements the request is approved and the vehicle is given a number plate as per 

the class of the vehicle which is unique and different from the other number plates. 

The registration process is robust but people with malicious intent have found ways of duplicating 

number plates. Duplicate number plates are either stolen from other registered vehicles or obtained 

from people who can create a resemblance of what is being used. Criminals are cloning number 



 

 

2 

 

plates and fraudulently registering cars (Stella Cherono, 2019).  These number plates are then 

patched on cars to make the vehicle look legal (Driver & Vehicle Licensing Agency, 2014). 

Duplication of number plates presents a major security threat, as criminals and terrorist tend to 

hide their identity during their criminal acts. Criminals and Terrorist understand that car number 

plates can be used to acquire information about the owner of the car and details about the car and 

hence the need to duplicate number plates. Duplicate number plates are either stolen from other 

registered cars, some are cloned and others fraudulently manufactured and registered. Up to the 

beginning of October 2018 there were already 22,487 recorded incidents of stolen number plates 

(Wheatstone, 2019).  

Vehicles with cloned number plate have become a common occurrence on the Kenyan roads, the 

vehicle is stolen, given an illegal number plate and some will go as far as cloning the vehicle with 

another vehicle that has more or less common features. This in turn will be used in covering up 

the identity of the vehicle in question and this will generate confusion with another existing car. 

In an article on shocking Kenyan cars with duplicate number plate published by the Youth Village 

Magazine, it described a victim who saw an exact car like his with the same number plates overtake 

him on Thika Road. Other cases reported by the daily nation on the terrorist attack on the 14th 

riverside attack revealed that the terrorist used a cloned vehicle. The Flying Squad had revealed 

that the number plate of the Toyota Ractis (KCN 340E) believed to have been used by the attackers 

was not a fake, but a duplicate of an existing one (Stella Cherono, 2019). 

To cab this issue of duplicate number plates automatic number plate technology in fixed location 

cameras is used. Automatic Number Plate Recognition (ANPR) within the UK is a powerful tool 

and its policing purpose is to deny criminals the use of our road (Rhead et al., 2012). Automatic 

number plate recognition technology makes use of the image recognition technology for extracting 

information out of an image and in this case, particularly the number plate of a vehicle. An 

algorithm will be used to get the details of the number plate from the image. Auto recognition of 

number plate method comprises of three segments: Character segmentation, Optical character 

recognition and template matching (Pechiammal & Renjith, 2018). 

Automatic Number plate recognition systems are used in parking systems to allow vehicles in and 

out of the parking zone also used in toll station by automatically detecting the number plate of the 

vehicle and therefore issuing a pay slip that will allow that particular vehicle to use that road 



 

 

3 

 

(Kashyap et al., 2018) In Kenya this system is being used in parking system and there is a thought 

of implementing the same in our toll stations along the major highways (Wang & Liu, 2016). 

According to the NTSA, the authority seeks to provide a seamless vehicle registration that 

integrates all the aspect of a vehicle without producing any duplications. Currently the transport 

sector relies heavily on paper-based/manual systems to provide services to citizens. There are 

different transport agencies and exchanging information amongst them is difficult with consequent 

losses and inefficiency. Similarly, it is difficult for transport sector institutions to integrate all 

existing applications while introducing new ones to meet the demand for providing better services 

to its citizens. This with other factors like corruption, theft of cars and manufacturing of illegal 

plates have led to the availability of duplicate number plates in our roads and some of these cars 

with duplicate number plates are used in illegal activities like theft and terrorism. With the 

availability of vehicles with duplicate number plates on the roads there is need to create a system 

that can be able to identify the duplicate number plate and flag them out while still on the move. 

The solution provided in this paper will entail using a system that can compare the details proceed 

from the image of a car that has been captured by different cameras in different location. The 

system will compare the second capture from the first capture putting into consideration the time 

one should take to move from the first point to the second point. If the time taken is smaller than 

the average time one should take, the details of the vehicle are flagged out. 

1.2 Problem Statement 

When it comes to duplicate and cloned number plates used by vehicles, the method being used is 

a reactive method whereby someone reports to the police of having seen a vehicle with the same 

number plate or when police get hold of a vehicle in a crime scene only to realise that the vehicle 

had cloned number plates after conducting a search with the vehicle registry. This method has 

proven not to be effective as it is usually a reactive response to the issue at hand. People don’t have 

to wait for someone to conduct a search to prove that there is a duplicate of the same. The research 

will focus on moving from the reactive response where it entails identifying a duplicate number 

plate posterior to a situation by making use of the cameras on our roads to identify the duplicates 

when the vehicles are on the move. 

The research intends to tackle the problem by designing a system that can detect vehicles with 

duplicate number plates and flag them out by use of the fixed location cameras on our roads.  



 

 

4 

 

1.3 Aim  

 The purpose of this study is to tackle the problem of illegal duplicate number plates by using the 

existing knowledge to come up with a duplicate number plate detection system using fixed location 

cameras.  

1.4 Specific Objectives 

i. To evaluate the current systems used in vehicle identification. 

ii. To analyse number plate identification using fixed location cameras. 

iii. To investigate the challenges associated with the current systems using image recognition 

to identify number plates. 

iv. To develop a duplicate number plate identification system using fixed location cameras in 

different locations. 

v. To test and validate the functionality of the proposed system. 

1.5 Research Questions 

i. What are the current systems used in vehicle identification? 

ii. How is number plate identification done using fixed location cameras? 

iii. What challenges are associated with the current number plate identification systems using 

image recognition? 

iv. How can a duplicate number plate identification system using fixed location cameras in 

different locations be developed? 

v. How will the functionality of the system be tested? 

1.6 Justification 

ANPR is an internationally recognized tool that is used in vehicle identification. ANPR systems 

allow for real time recognition of a vehicle’s number plate thus saving the time spent in vehicle 

identification in certain instances (Simon et al., 2017). 

Vehicles involved in crime usually have a cloned number plate or a duplicate number plate. A 

system that flag’s out vehicles with duplicate number plate when they are on the move helps reduce 

crime and acts as a deterrent measure when it comes to terrorism. This system does save the time 

used in proving that a vehicle is duplicate. From Figure 1.1 it shows a graph of the cases reported 

on duplicate number plate in the year 2017/18 and 2018/19. You can easily tell that the cases being 

reported are on the rise and hence a need for a system that is able to identify duplicate number 



 

 

5 

 

plate system. This research will aim at coming up with a system that is able to flag out duplicate 

number plates. This research will assist mostly the people in the security sector as they will be able 

to know or have information of the vehicles with duplicate number plates. 

 

Figure 1.1: Car Cloning Report (Wheatstone, 2019). 

1.7 Scope and Limitation 

This research focuses on a system that is used to detect duplicate number plates of vehicles. With 

the use of cameras, the system is able to extract the number plates give time and location the image 

was captured then compare with what is stored in the database to determine if there is a duplicate 

of the same number plate. The study focuses on use of image recognition to identify number plates 

and google locations to give distance and the average time one should take to move from the first 

camera to the second camera. 

 

  



 

 

6 

 

Chapter 2: Literature Review 

2.1 Introduction 

This chapter reviews relevant literature to further enhance the understanding of the features used 

in vehicle identification and ways of identifying number plates. Discussion on what it entails to 

flag out a duplicate number plate, challenges faced by the current methods being used in vehicle 

and number plate identification and how it is done are detailed in this chapter. A conceptual 

framework is then presented at the end of the literature review to describe how the researcher 

intends to approach the issue of identifying duplicate number plate by use of fixed location 

cameras. 

2.2 Vehicle Identification 

Vehicle detection and identification techniques have been widely applied to acquire vehicle 

information depending on various sensors (C. Zhang et al., 2017). Vehicles can be similar in shape, 

colour and model but each vehicle can be uniquely identified by its plate number in combination 

with specific features like the shape, colour and model of the vehicle. Works have already been 

done describing the various methodologies used in vehicle identification and extraction of vehicle 

information. 

Vehicle identification is a primary requirement for Intelligent Transport Systems (ITS), these days 

three common techniques are available (Al-bakry et al., 2017). The three methods mentioned by 

Al-bakry and Mushatet include use of automated number plate recognition system (ANRP), that 

allows users, to automatically monitor vehicles by use of cameras to extract their number plates. 

Second method entails use of the barcode system to identify the vehicle and the last method used 

in vehicle identification is the use of the radio frequency technology (RFID). 

The most common method of vehicle identification is the use of computer vision to identify vehicle 

number plate. Computer vision is the transformation of data from a still camera or video camera 

into a new decision or a new representation (Suryatali & Dharmadhikari, 2015). The use of 

computer vision to derive number plates from the images led to development of the automatic 

number plate recognition from vehicle images. This automatic number plate recognition system 

(ANRP) was originated in 1976 at the police scientific development branch in UK , it is a mass 

observation method that uses optical character recognition on image to read vehicles registration 



 

 

7 

 

plates (Al-Bakry et al., 2017). ANPR consist of a camera that is linked to a computer. When a 

vehicle passes by the camera, the camera records an image which is automatically ‘read’ by the 

computer and the vehicle registration mark (VRM) recorded (NPCC, 2016). A national police fact 

sheet published by the UK National Police Chiefs Council in April 2016 states that the use of this 

ANRP systems is used to support operational responses, information and intelligence and 

investigations. In operational responses the, vehicle captured by the ANRP camera registration 

number is identified and checked against the database records of vehicles of interest and if the 

vehicle is listed as a vehicle of interest the police are able to intercept and they can make an arrest 

if need be. 

The police services require intelligence on movement of vehicles on the road and through the 

ANRP camera systems they are able to monitor what is happening on the roads and gather specific 

details of the vehicle from the systems. The use of ANPR has been of important in the detection 

of many offences, including locating stolen vehicles, tackling uninsured vehicle use and solving 

cases of terrorism, major and organised crime (NPCC, 2016). 

With the use of fixed location cameras in different locations, ANRP will be used to identify vehicle 

number plate and record other details like the time of capture and the location of capture.  In 

Bengali the ANRP has been implemented in such a way that it does not only give information on 

the vehicles but an algorithm is added to it in order to assist the system in traffic management in 

metropolitan and mega cities (Shahed et al., 2017). 

Apart from ANRP we have the Barcode technology which is the oldest but it is rarely used as 

vehicle identification (Al-Bakry et al., 2017). Barcode refers to a method of storing data using 

geometric figures and in this case the data being stored is the vehicle details. The quick response 

(QR) code which is majorly used in vehicles is a barcode that has two-dimensional barcode and 

was developed in Japan by Denso Wave Cooperation. The information related to that vehicle is 

embedded on the barcode and there are barcode scanners that are used to retrieve the information 

(Suryatali & Dharmadhikari, 2015). Vehicles just like any other objects that are usually stationary 

can be identified by use of barcodes and QR codes. A bar code reader is usually mounted along a 

street whereby its able to read the bar code and retrieve the information. 



 

 

8 

 

Another technology used to identify vehicles is the Radio-Frequency Identification (RFID). RFID 

is the investment of radio waves in reading and capturing information stored on a tag tied to an 

object (Al-Bakry et al., 2017). RFID is a technology that provides wireless identification of a 

vehicle and is more robust than a bar code (Balbin et al., 2018). In this method of vehicle 

identification two things are involved; RFID tag which usually is placed on the windshield of the 

vehicle and has the vehicle details. The other element is the RFID reader which is usually in a 

fixed position and should be able to communicate with the RFID tag in order to produce the vehicle 

details. 

2.3 Number Plate Identification Systems 

Number plate identification is the process of pinpointing vehicles by their number plates.  Each 

vehicle upon registration in a country, is issued with a unique identification number also known 

as the registration number. This registration number is made up of characters that are used in 

identification of the vehicle and through this you can tell who owns the car, model and year of 

manufacturing of the car. Systems used in vehicle identification will focus mostly on identifying 

the unique registration number. To avoid identification of the vehicles, number plates are usually 

cloned, and the cases of cloned number plates is on the rise. To aid in duplicate number plate 

identification we need to understand identification of this number plates There are various ways 

that can be used in number plate identification. The methods are also commonly used in securing 

number plates as well as explained by Hoffman (2013) on his article on securing number plate 

based on digital signatures and radio frequency identification (RFID).  

2.3.1 Security Marking Technologies 

This is not a common technology, but it applies the same principle of security markings in bank 

notes used to avoid counterfeiting (Hoffman et al., 2013). According to the Computer Resource 

Security Centre, security markings refers to the means used to associate a set of security attributes 

with objects in a human-readable form. It includes use of holograms watermarks as well as security 

inks. Identification tags are coded with a given number, or for use on vehicles,  asset identification 

number or a unique serial number (Data Dot Technology USA[DDTU], 2018.). The security 

marking technologies provides a unique security code for you. This unique code is usually 

registered to the owner of the marked vehicle number plate. Thus once the code on the number 

plate is detected it gives information on the ownership of the vehicle. 



 

 

9 

 

2.3.2 Advanced Electronic Signature 

This is advance of the security markings and it makes good use of cryptography for data security. 

Cryptography brings in the mathematical algorithms and digital sequence to transform data in such 

a way that for someone to access the data he/she must use the correct key. The nature of the 

mathematical algorithm is intentionally chosen to make it as difficult as possible to derive the 

cryptographic key from any number of encrypted samples (Hoffman et al., 2013). Cryptography, 

encryption techniques based on mathematical algorithms are used for the creation and verification 

of the digital signatures (Sreeja & Misbahuddin, 2018). Asymmetric encryption differs from 

symmetric encryption in the sense that a set of two keys is always used: one key (normally called 

the Private Key) is used for encryption, while another matching key (called the Public Key) is used 

for decryption of that data (or vice versa). 

Validation of the digital signature on the number plate will include comparison between the two, 

which one is a human readable content and the other is a digital signature, and by recording each 

digital signature while issuing a number plate will establish a reference of comparison, which can 

be used to detect illegal, or legally issued number plates. The disadvantage with this system is that 

a criminal will try to use a number plate that was legally registered. Another disadvantage with 

this system is that digital signatures in the form of private keys and public keys are used to validate 

files sent. If it is validated that there is a fake signature by a user, then the user cannot deny it 

(Setiawan & Rey Citra, 2018). Figure 2.1 shows the flow of a digital signature process. 



 

 

10 

 

 
Figure 2.1: Digital Signature Process Flow (Hoffman et al., 2013). 

From the diagram above the advanced digital signature is composed of mainly three elements 

whereby we have the first element being the description unit, whose main function is to generate 

the digital signature. After this the digital signature is generated and the third function entails 

packaging the data to be stored in a medium. The description unit is made up of the data attributes 

which is information about the data. After defining the data attributes, the data is encoded into a 

digital signature. The encoded medium or data is then attached into a number plate as shown in 

the above diagram. 

2.3.3 2D Barcoding 

Barcoding, a system of labelling using spaces and bars printed side by side on an object, was the 

first widely used technology for the automated identification of digital codes (Hoffman et al., 

2013). The digital codes are used to store the information and this information is retrieved via a 

bar code reader. The disadvantage with this is that the barcode can be copied, printed and used in 

a different number plate thus making it difficult to tell an illegal or legal number plate. 2D 

barcodes, where the pattern is made up of blocks or dots, can store up to several kilobits making it 



 

 

11 

 

possible to use them as carriers of digital signatures (Hoffman et al., 2013). The figure 2.2 shows 

an example of a 2d bar code image. 

 

Figure 2.2: 2D Barcoding (White & Gardiner, 2007). 

The code for the barcode is usually generated and then evaluated. The message is a randomly 

generated 662-bit sequence and it is encoded with a 1324-bit code word of a (3,6) LDPC code (L. 

Zhang et al., 2019). The coded message is then one that has the information or the vehicle details.  

2.3.4 Radio Frequency Identification (RFID) 

Radio frequency identification is a form of a wireless communication that integrates electrostatic 

coupling principles in the radio frequency spectrum to identify an object, animal or a person. Radio 

Frequency identification (RFID) is a technology that provides wireless identification and tracking 

capability and is more robust than a bar code (Balbin et al., 2018). A sticker is put on the vehicles 

number plate metal or vehicle wind shield and an RFID reader should be able to read the sticker 

in order to get the information from the sticker. The information contained in the sticker is usually 

the person registered as the owner of the vehicle, model, colour, type and date of registration. This 

has an advantage over barcode, as the vehicle details can be obtained while the car is still in motion 

unlike in barcode reader the vehicle should be static and at a close range.  

A reader uses its own antenna to communicate with the tag. When a reader broadcasts radio waves, 

all tags designated to respond to that frequency and within range will respond (Suryatali & 

Dharmadhikari, 2015). The RFID scanner does not have to be in the same line of sight with the 

tag if it’s on the same wavelength. The passive RFID has a capability of reading and processing 

multiple tags at the same time. 



 

 

12 

 

RFID tags can only hold about 1 Kilobyte (8 Kilobit) non-volatile EEPROM storage and 4-byte 

unique identifier burned into the chip just enough to store an identification number. There are two 

main types of RFID: active (where the tag is battery operated) and passive (where the tag is 

powered by the reader) (Hoffman et al., 2013). The battery in the active RFID makes it expensive 

compared to the passive and this enables the passive RFID to be massively produced and used 

compared to the active one. 

 
Figure 2.3: Components of an RFID System (Jia et al., 2012). 

From Fig 2.3 we have the tags which are also referred to as the transponders and are always 

attached to the object in our case it’s the wind shield of the vehicle. Tags consist mainly of a coiled 

antenna and a microchip, with the main purpose of storing data (Jia et al., 2012). The second 

component is the reader which is a radio frequency unit that is to initiate communication with the 

tag and ensure that data is also retrieved from the tag. Figure 2.4 below demonstrates a 

communication between the reader and the tag. The diagrams describe how the packets move from 

the initiator of communication to the last point. 



 

 

13 

 

 

Figure 2.4:Reader and Tag Message Communication (Larionov et al., 2017). 

2.3.5 Automatic Number plate recognition. 

ANRP uses sensors like cameras to capture the image, which is then manipulated to detect the 

region of interest, the image is then normalised, and optical character recognition technique is 

applied to retrieve the number plate. This is by far the most common used method in number plate 

recognition but in this case it’s not a complete system because the idea is to not only identify the 

number plates but to identify duplicate number plates and flag them out. The problem with this is 

that there is no means to prove that a number plate is legal or not. The Automatic Number Plate 

Recognition (ANPR) system for vehicle identification was invented in the year of 1976 at the 



 

 

14 

 

Police Scientific Development Branch in the United Kingdom (Babbar et al., 2018). However, 

there has been a steady growth for the last decade due to the improvement on the digital cameras. 

As stated by Mutua, S.M in his research paper on an Automatic Number Plate Recognition System 

for Car Park Management System an ANRP is composed of a camera, computer, software to detect 

the license and the database. The camera is the physical device mandated in image collection and 

the computer is where the license plate recognition software is housed. The database is used to 

store the number plate identified and any other data that might be collected by the camera. 

Bandar Sunway in his paper on Automatic Vehicle Number Plate Recognition Using Structured 

Elements he proposed an algorithm of automatic number plate recognition which entails the 

following steps: Taking picture from the vehicle, processing the image, character extraction from 

the number plate, character segmentation, character matching, identify the number plate, output. 

Once the image is captured, we start by converting the image into a grey scale image. Greyscale 

Images are those images which contain only a single value that is each pixel has only a single 

value, they carry only the information of intensity under them (Kashyap et al., 2018). Figure 2.5 

below shows an image capture of a vehicle from a camera. 

 

Figure 2.5: Image Capture from a Camera (Dias et al., 2019). 

The image is then converted into a Binary image. A binary image is an image whose pixels have 

only two possible intensity values. They are usually displayed as black and white. Normally, the 



 

 

15 

 

two values are 0 for black, and 1 for white (Kurdi & Ahmad, 2017). Threshold is usually used in 

creating a binary image. A vertical edge detection algorithm is used to detect the edges of the 

number plate. The accuracy rate of this technique to identify edges is 92.5% (Saleem et al., 2016).  

The figure 2.6 below shows a sample of a Binary image after the threshold process. 

 

Figure 2.6: Binary Images (Kashyap et al., 2018). 

In the last two steps it entails separation of characters and only figures are left for identification or 

recognition. In this case Artificial Neural Network is applied in order to identify the figures. 

Artificial Neural Networks are vastly used intelligent calculating design for recognition of patterns 

(Kashyap et al., 2018). Figure 2.7 summarizes all the process involved in number plate 

identification from an image. 



 

 

16 

 

 

Figure 2.7: Steps for Numbers Plate Identification From an Image (Kurdi & Ahmad, 2017). 

2.4 Machine Learning to Identify the Number Plates 

Here, we will look at the various ways and frameworks used in identifying number plates from 

images. Machine learning can be defined as a set of methods that can automatically detect patterns 

in data and then use the uncovered patterns to predict future data or perform decision making under 

uncertainty (Robert, 2014). Machine learning technology is usually used to identify objects in 

images, transcribe speech into text, select relevant results but in our case, we will focus mostly on 

object identification in images. Image recognition in the context of computer technology is the 

aspect of identifying objects, people places and texts on images by use of a software. 

2.4.1 Image Acquisition and Pre-process  

High-end IR cameras, placed at strategic positions to avoid obstacles in order to obtain accurate 

images, perform image acquisition.  Image acquired by camera always reflect the camera settings; 

among many include colour and hue, saturation and value or brightness, whereby essentially an 

image can be in its natural form or slightly altered. Colour images are complex in space and time. 

Pre-processing aims at image enhancement and restoration. This process eliminates noise, 



 

 

17 

 

highlight edges and improve the overall quality of an image. In image pre-processing, an image 

goes through among many procedures grey scaling, dilation, erosion, and filtering and edge 

enhancement. Converting natural colour images, to high saturation value colour space, to grey 

scale and then to binary is important as it reduces time and space complexities. 

The image acquired should be of good quality as this will determine or affect the recognition result. 

Sensors like cameras will be used to identify the presence of a vehicle and if a vehicle is detected 

an image of the vehicle is captured. The camera should be of high definition and after the images 

are captured, they are stored in a database. 

Image processing is a technique where by an image is converted into a digital form and various 

operations are performed on this image in order to get an enhanced image to extract information 

(Pervez, 2013). Below see the details of the various operations performed. 

(i) The image is imported by use of a scanner or digital photography 

(ii) Image manipulation is done to enhance the image quality as well as compress it. 

(iii)In this operation stage analysis is done on the enhanced image to retrieve the information. 

In this last stage where the analysis is done on the image Convolution Neural Network (CNN) is 

used.  

 

2.4.2 Number Plate Detection and Extraction  

Automatic Number Plate Recognition (ANPR) is a kind of image processing technology for 

recognizing the vehicle number plate. This system also offers users to place mark out and monitor 

moving vehicles automatically by extracting their number plates (Islam et al., 2016). This type of 

technology was first introduced in 1979 in the United Kingdom. Automatic number plate 

recognition systems have various applications that include traffic monitoring, parking lots control 

and identification of plundered vehicles. 

Therefore, in order to have reliable number plate detection algorithms, in such dynamic 

environments, several choices must be considered. Several features that could be deployed to 

extract a rectangular shape of number plate from a car image included colour feature, aspect ratio, 

and texture edge density and shape/size but for better detection rate the combination of features 

could provide more reliable solution. Hue saturation value colour space and integral image 

properties could also be employed to locate the coordinates and position of yellow number plate 

and non-yellow number plates. From Yogheedha (2018) HSV enables one to find out the four 



 

 

18 

 

coordinates of a rectangular shape containing English like symbols or texture from which X and 

Y coordinates, Width and Height could be extracted. The process of number plate area detection 

and extraction involved three stages of finding area of interest, filtering out background and 

removals of false objects, and lastly computation of connected components, which provided best 

results of required region of interest.  

2.4.3 Character Extraction and Segmentation  

Image segmentation is usually the first or initial stage of image processing(Saini & Arora, 2014). 

Segmentation entails locating object and boundaries in an image. This has been made possible 

through computer visions whereby models are able to detect objects, determine their shape, give 

prediction on the direction the objects will move in and many other things. In this case 

segmentation will be used to locate the position of the number plate and the characters labelled on 

the number plate. Basically, there are two image segmentation techniques: Edge based and region-

based technique. Edge based is segmentation method based on discontinuity find for abrupt 

changes in the intensity value(Saini & Arora, 2014). Segmentation by region based relating the 

pixel and grouping them(Nikhil & Sankar, 1993) 

It is must to recognize the vehicle number accurately, which is mostly dependent on the character 

separation or isolation. So all the character from the image are separated without losing any 

element of a character (Islam et al., 2016). Before you introduce the character recognition 

algorithm the image noise needs to be reduced by a process called normalization. Normalization 

is to refine the characters into a block containing no extra white spaces (pixels) in all the four sides 

of the characters. Then each character is fit to equal size (Ozbay & Ercelebi, 2005) 

Some previous researchers who contributed to recognition and interpretations of characters 

proposed methods like Markov model, neural network approaches, statistical pattern recognition, 

chain code, template matching and many others. Each of these methods has advantages and 

disadvantages dependent on the working environments as well as the weather conditions. This was 

part of this research; the processor to recognize snapshot captured by camera and return a text 

representation of the detected image. 

2.4.4 Vehicle Colour Recognition 

Among the many recognition features of a vehicle colour is the most discernible feature and easy 

to recognize to the human eye. The image acquired by the camera be it of the front or rear end of 



 

 

19 

 

the car can be used to identify the colour of a particular vehicle. The colour identification of 

vehicles plays a significant role tracking crime vehicles in imaging systems based on CCTV (Jeong 

et al., 2018). The colour of a vehicle is a very important clue as most witnesses in a crime scene 

will easily recognize the colour of a vehicle and not the number plate hence this is a very important 

feature that should be taken into consideration. Due to the mentioned reason, it is of importance if 

we can be able to identify the colour of the vehicle as well. 

Vehicle colour recognition is one of the effective ways of vehicle recognition. Some researchers 

have proposed a colour recognition system using support vector machines (SVM) and 

convolutional neural network (CNN) which is good in rich colours but this method does not 

include white and silver colour classification (Tilakaratna et al., 2017). The rationale behind the 

exclusion of the white and silver colour in the classification scheme is because the two have pixel 

values that are close to each other thus leaving no room for single colour space. The system will 

try to classify up to thirteen colours, which are white, grey, silver, black, sky blue, red, blue, brown, 

orange, pink, green, yellow and finally purple. 

After the image has been captured, the system will identify a region of interest just slightly above 

the number plates of the vehicle. To successfully implement the system and be able to classify 

thirteen colours the system implemented two main methods for colour recognition. 

(i) Histogram and back projection method – it’s a technique used in finding objects of interest 

in an image. This method was implemented without any machine learning method. The 

model for each colour was found after doing some research on colour values of vehicle 

images (Tilakaratna et al., 2017). Back projection is a way of recording how well the pixel 

of a given image fit the distribution of the pixel. The system will calculate the histogram 

model for region of interest and then uses it to find its feature in the image. 

(ii) Support vector machines (SVM) based method. 

A novel colour correction technique will be used to reduce the image changes that will affect the 

quality of the result so that the vehicles can be more accurately identified. Colour histogram feature 

is a one of good index to analyse the colour variation and the colour distribution in an image (Jeong 

et al., 2018). In Yoosoo Jeong research paper on homogeneity, patch search method for efficient 



 

 

20 

 

vehicle colour classification using front-of-vehicle image he proposes the use of search method of 

homogeneity patches to segment the reference region for colour classification. 

In computer vision, we have several models used for describing the specifications of the colours 

the models include  

The first phase in this method will entail identifying the region of interest to execute the 

homogeneity patches searching method and the Hue Saturation Value histogram feature is 

extracted for each value. After this, the multicast ad boost algorithm with colour features is applied. 

2.5 Challenges Associated with Current Automatic Number Plate Recognition 

Systems 

Table 2.1 shows a summary of the challenges faced by the systems used in automatic number plate 

recognition. 

Table 2.1: Numbers Plate Detection Methods. 

Approach Merits Demerits 

Security Marking 

Technology (Balbin et 

al., 2018) 

Robustness: should be 

able to with stand 

normal signal 

processing such as 

image cropping. 

Anyone who has access to the technology 

can be ab 

 

le to create cloned number plates. 

Advanced Electronic 

Signature (Balbin et al., 

2018) 

It is unforgeable. Criminals will try to use a number plate that 

was registered. 

Automatic number 

plate recognition 

system (Pechiammal & 

Renjith, 2018) 

Can be applied in 

different systems and 

can be used while the 

vehicle is still in 

motion. 

With this system, it is difficult to prove if the 

number plate is legal because it could have 

been fake registration. 



 

 

21 

 

2D Barcoding Balbin et 

al., 2018) 

Fault tolerance: it can 

still restore 

information even when 

slightly damaged. 

Difficult to use this system while the vehicle 

is still in motion. 

Radio Frequency 

Identification 

(Larionov et al., 2017) 

Can be used while the 

vehicle is in motion 

They are less reliable compared to the other 

systems 

2.6 Conceptual Model 

From this literature, the researcher has conceptualized the use of an expert system to aid in flagging 

out of duplicate number plates while the vehicles are on the move. The researcher work with 

creating a system that can receive captured images by the fixed location cameras. This system will 

have a central database to record the vehicle number plate color and the time the vehicle was 

captured by the camera. The researcher will use automatic number plate recognition system and 

histogram back projection system to identify the number plate and color of the vehicle. Time is 

the key facture because it will be used to determine if the vehicle captured for the second time is a 

duplicate. For a vehicle that has been captured by camera A to move to Camera B considering that 

it is moving with 110km/h it should take X time. If the time taken to move from camera A to 

camera B is less than X then we consider this to be a case to be flagged out.  

 



 

 

22 

 

 

Figure 2.8: Conceptual Model. 

  



 

 

23 

 

Chapter 3: Research Methodology 

3.1 Introduction 

Research is all about innovation and ingenuity. The initial step being to decide the core problem 

that must be solved. After which comes the part where a meticulous approach has to be applied 

with due diligence to come build an approach roadmap that leads to the solution of the 

aforementioned problem or challenge (Soni & Bhadauria, 2019). The problem at hand is how to 

identify duplicate number plates by use of fixed location cameras. 

This chapter is all about research methodology which will entail clearly defining road map that the 

researcher has followed to provide an amicable solution to the identification of duplicate number 

plates by use of fixed location cameras. The researcher clearly defines the road map that is being 

used to come up with a system that identifies duplicate number plates and flags them out. Finally, 

this chapter does give details on any concerns or issues that may arise due to the ethical quality of 

the work.  

3.2 Research Design 

A research design is a plan according to which one obtains research participants and collects 

information from them (Mugenda & Mugenda, 2003). The research design refers to the overall 

strategy taken to integrate the different components of the study in a coherent and logical way, 

thereby, ensuring that the issue of identifying duplicate number plates while vehicles are on the 

move is effectively addressed.  

This constitutes creating a system that is able to compare inputs from the available fixed location 

cameras with respect to time distance and speed in order to be able to flag out duplicate number 

plates. The research design adopted in this project is an action-based research that aims to come 

up with a solution to the challenges faced by the current vehicle recognition systems. 

3.3 System Development Methodology 

The approach that has been used is the prototyping approach, which involves building an early 

prototype and then testing, and reworking on it until an acceptable product is achieved. The 

prototyping approach that is being used is the Extreme prototyping approach. The system 

development methodologies have evolved over time, each with its own advantages and  



 

 

24 

 

disadvantages. One of the widely cited advantages of using methodologies to systems development 

industry is that it makes the development process more genial to project management, monitoring 

and control (Ally & Ning, 2015).  

3.3.1 Prototyping Methodology 

Prototyping is a methodology that evolved out of the need to better define specifications and it 

entails building a demo version of the software product that includes the critical functionality (D. 

S. Journal, 2014). This methodology will assist in coming up with a system that is a prototype and 

then expose the system to users for comments so at to rework it and ensure an acceptable system 

is developed. Figure 3.1 shows the different development stages that has been incorporated in the 

study for the development of the system. This prototyping model is often considered as the 

extension of waterfall. 



 

 

25 

 

 

Figure 3.1:Prototyping Development Model (Madhav mohan, 2019). 

(i) Prototype development 

Prototyping is a methodology that evolved out of the need to better define specifications and it 

entails building a demo version of the software product that includes the critical functionality. This 

phase involves gathering the requirements for the system. Studying already existing similar 

systems in order to formulate an overview of what functionalities the system should have. Based 

on the gathered requirements a design is made to exhibit the functionalities of the system. This 

design will be used to educate anyone who is interacting with the fixed location cameras or the 

parties involved in identification of duplicate number plate which happens to be the security 



 

 

26 

 

personnel. If the client is not pleased with the system, the additional requirements are put into 

consideration and another prototype is built. Once the client approves of the prototype it is then 

discarded and the experience gained in building the prototype is used in the next phase. 

(ii) Design Stage 

This will assist in specifying the hardware required and provide an overview of the system being 

developed. This helped in designing a system that is compatible with any fixed location cameras 

used for the transport management system. At this stage the design is documented as it will guide 

the researcher in development of the system to flag out duplicate number plates. 

(iii) Coding 

After all the analysis and design, the actual coding of the system begins and implemented into the 

initial iteration of the project. Here we will see the design of the system being developed into an 

actual system. If there is need be this methodology allows the researcher to go back and review 

with the clients. 

(iv)  Testing 

Upon completing the implementation process, the system is to be tested to identify any issues or 

potential bugs. The functionalities of the proposed system will be tested to the full. The efficiency 

and accuracy of the system to identify duplicate number plate is put to the test here. The researcher 

checks if the system can meet the recommendations that are required. After the system is tested it 

is again given back to the clients for evaluation. 

(v) Maintenance 

There are some concerns which come up in the client atmosphere. To repair those issues patches 

are released. Also, to enhance the artefact some better versions are released (I. Journal & 

Engineering, 2015). This is usually done to bring the changes to the client and help on improving 

the model. In case the system will require an improvement then this can be an add on during the 

maintenance process. 



 

 

27 

 

3.4 Data Collection Instruments 

During the research secondary data sources were used. The experimentations allowed the 

researcher to find the best datasets which had different sets of number plates on different sizes as 

well different car models. The data that will be used will be obtained from publicly available 

Kaggle database. 

Availability of the secondary sources most of which have been discussed in chapter two enabled 

the researcher to compare different algorithms and different application programming interface 

used in extracting the car model, number plate and the colour of the vehicle Additionally, these 

sources also assisted the developer realize the best open-source software to use to develop the 

system’s prototype. 

The proposed model used base training data as a base comparison to the input data. From the input 

data the system was able to draw conclusion. In the performance and data analysis fifty car images 

were used to test the system and the results were successful as the system was able to identify and 

hence achieving an accuracy of 99% and above. 

3.5 Database Development Tools 

The reason for using MySQL is that it is an open-source database that is easily connected to other 

software’s and is also able to provide optimum efficiency even when working with some of the 

most demanding applications. MySQL has a way of automating everything from the expansion 

and configuration of data design and database administration. Its security and high-performance 

nature make it more viable for even large project that require privatization of data and hence 

making it the best for the storage of the images captured and also storing data retrieved from the 

images. 

3.6 Programming Tools 

The reason for using python programming language is that it is very simple and fast to implement. 

In addition, many community-based resources support python, especially for data analysis, which 

is a core part of this research. 



 

 

28 

 

The system will three major modules which will include the front end, the back end and the API 

used to retrieve the make model and number plate of vehicles. The languages used in the 

implementation of the different system modules are as stated below. 

(i) Backend 

This is the module that ensures that the client end is working by storing and organizing data, It 

ensures that the front end works by sharing or sending and receiving information to be stored on 

the front end. The language to be used here will be Laravel PHP framework. The beauty about 

Laravel is that it’s a lightweight and can be used seamlessly with databases enabling the query of 

database using the PHP syntax. 

(ii) Frontend 

This is basically what the user or client interacts with, it’s usually accessible from the browser. It 

includes everything the user interacts with or experience’s, including the colors texts, images and 

navigation buttons or menus. This module will be created using the Android Java Min SDK 

Lollipop (API 22). This is because the researcher will require to use the phone camera in place of 

the actual road cameras. 

3.7 System Analysis  

As part of the system analysis, this research will seek to do the following: Determine who are the 

users of the system. The goal here is to make sure that the system developed will be of as the 

requirements of the users. If wrong users are identified, there is a possibility of developing the 

wrong solution. The users for the system developed in this research will be the security personnel.  

3.8 Research Quality 

The research data must have both validity and reliability. Validity is the extent to which the data 

accurately measures what they were intended to measure, whereas reliability is the extent to which 

the data collection method will yield the same findings if replicated by others. The research quality 

will be determined by the validity of the tools being used and by checking if all the research 

questions have been answered. The researcher will ensure that all work used in this research will 

be well cited and any work used in the literature review will be from reputable sources 



 

 

29 

 

3.9 Ethical Considerations  

 Ethics are standards or codes of conduct that help distinguish between what is acceptable and what 

is not (Ethical Considerations: CIRT, 2018). The ethical consideration in this research will be 

determining the truthfulness of the data obtained from the secondary sources which is an open-

source public database called Kaggle. All previous works were cited appropriately, and due 

acknowledgement given to the respective authors. 



 

 

30 

 

Chapter 4: System Design and Architecture 

4.1 Introduction 

In this section the researcher gives a layout of the design architecture of the developed automatic 

number plate recognition system. The architecture will be based on the conceptual model diagram 

outlined in chapter 2. The design architecture plays a critical role in the development process as it 

provides a well laid out structure on how to come up with system. The researcher intends to use 

class diagrams, use case diagrams, data flow diagrams and data flow diagrams to show the 

interaction of the user, developed system and the components of the developed system. 

4.2 Requirement Analysis 

Requirement analysis will assist in determining the feature expectations as well as the functional 

specifications by breaking it down into smaller steps. In order to develop the system, the researcher 

categorized the requirements gathered into two categories: the functional requirement and non-

functional requirement. 

4.2.1 Functional Requirements 

Functional requirements describe the specific activities to be performed by the system. The 

functional requirements for the system as identified by the researcher include. 

i. The System should be able to accept a series of images from the Cameras. 

ii. The system should be able to tell the time and location of the image captured by the camera. 

iii. The System should be able to extract the relevant information from the images which is 

vehicle number plate, color, and model. 

iv. The system should be able to record the extracted information from the image. 

v. The system should be able to compare any new entry with what has been recorded before. 

vi. The system should be able to give an output after comparison. 

vii. The system should send a notification if there is a duplicate number plate that has been 

recorded. 

4.2.2 Non-functional Requirements 

Non-functional requirements are used to set the quality attributes of a software; these attributes 

will be used to determine the quality of the software. 



 

 

31 

 

4.2.2.1 Supportability Requirements 

The System should be able to fit in a wide range of cameras used on the roads and can be easily 

used to provide the desired output. This will provide flexibility in terms of the different cameras 

being used the system will function as it should. 

4.2.2.2 Reliability Requirements 

The system should firmly provide the same results with the same input data. The system should be 

able to give the details of the captured image with precise accuracy to avoid false recording of 

information in the system. In case of a failure the administrator should be able to restore the system 

to its functionality. 

4.2.2.3 Security Requirements 

Software security is to engineer a software in such a way that the required application functions 

uninterrupted and can handle nicely the security threats during a malicious attack (Daud, 2010). 

The system needs to have strong security measures to ensure that integrity and privacy of 

information is protected. 

4.2.2.4 Scalability Requirements 

If there is an increase in the number of images being captured and information stored the system 

should be able to handle the extra load without failure. The system should be able to capture more 

images and be able to give a complete comparison and give feedback. The system also deletes 

images stored for more than six hours in the system. 

4.3 System Design and Architecture 

The system will be using Images captured by the cameras as the only input data. The image 

captured is then passed on to the system where image processing is done, number plate detection 

as well as the character recognition in order to give the identified number plate, colour and model 

as the output. The system is also designed to give the time, the image is captured as well as the 

location of the camera by use of latitude and longitudes of the camera location. Before storage of 

the data in the database a comparison is done to establish if there is such information and if so a 

comparison on the distance if there is the time in order to flag out cases where we have 

inconsistency in terms of time and distance. This well detailed in the Figure 4.1: 



 

 

32 

 

 

Figure 4.1: System Architecture. 

4.3.1 Data Input 

In the proposed system, the Image input is done at this level after the image of a vehicle has been 

captured by a camera in a specific location. The Image location and time of capture are the details 

to be captured or recorded in the database. The camera should be able to deliver an image that can 

be analysed even in different weather conditions especially when it’s raining and the visibility is 

affected. 



 

 

33 

 

4.3.2 Data Processing 

The stored images are then sent to an application programming interface that is connected to the 

system to do the image analysis. From this, the output is the number plate, model, and colour of 

the vehicle. This is then compared to what is in the database and there is no match of the number 

plate the system stores the data in the database. If there is a match, a time and location comparison 

is done to establish if it’s a duplicate. If the time taken by the vehicle to move from camera A to 

camera B is within the threshold, then it’s not considered as a duplicate. If the time taken for the 

vehicle is less than the given threshold, then it is flagged out as a duplicate. 

4.3.3 Classification Output 

This is the final stage of the proposed duplicate number plate identification system. The output of 

this is only generated when we have a duplicate that has been identified and it’s displayed on the 

system as a duplicate number plate. This will be important information especially to the security 

personnel as they can start tracking the vehicle since they have the details. 

4.3.4 Process Design 

This section outlays the set of input resources which have been used to transform the input into 

the meaningful output information. The system has used different languages to make sure that the 

system processes and gives the relevant output and reflects the actual thing. A detailed information 

of this section has been discussed in chapter five under software development environment. 

4.4 System Design 

The collection of the different objectives as from the analysis of the literature and other similar 

systems were merged with the ideas that the developer had in mind to synergize an application 

design with desirable functionalities to fulfil its objectives. The following design diagrams has 

been used to give an insight into the actual implementation of the system. This includes the 

different system’s modules. 

4.5 Data Flow Diagram 

Data Flow Diagram (DFD) provides a visual representation of the flow of information that is data 

within a system. In the proposed system, this diagram has been used to show the information 

provided by and delivered to someone who takes part in the system process, the information needed 

to be stored and accessed.  



 

 

34 

 

Context Diagram 

Context diagram gives only the top-level characteristics of the system data flow. It demonstrates 

only one visible process node that represents the functions of a complete system in regard to how 

it interacts with external entities. It therefore briefly shows the boundaries of the proposed system 

thus viewing the system as a black box.  

 

Figure 4.2: Data Flow Diagram. 

Level 0 Diagram 

The different entities initially in the system as a black box have been broken down into various 

processes which interact with the external entities. The data stores represent the storages done in 

the system during processing. Some of the data stores are temporary as they involve temporary 

storage of information before processing or display.  



 

 

35 

 

 

Figure 4.3: Level 0 Diagram. 

4.6 Use-Case Diagram 

A use case is a list of actions and event steps that define the interactions between an actor and a 

system to achieve a goal. In the proposed model, different cases have been simulated 

diagrammatically as shown in the figure 4.4. The entities involved included the independent 

system classes such as the classier of the number plates function that is in the system but interacts 

with the system components. 

Some of the assumptions that has been made in the use case diagram includes. 

i. The display and the client machine will do the actions independently even though in 

the system, the machine responsible for both actions is one. 



 

 

36 

 

ii. The camera auto controls itself to capture the image and sends the image to the database 

then resets to capture another image. 

 

 

Figure 4.4: Use Case Diagram. 

 

4.6.1 Detailed description of Use-Case Diagram 

This section provides a comprehensive description of the use cases as shown by the Table 4.1 

below. 

 



 

 

37 

 

Table 4.1: Use Case Diagram. 

Use Case  Data Pre-processing 

Primary Actors Fixed Location Camera 

Pre-condition Camera is on and connected to the system 

Post-condition Vehicle image capture 

Main Success Scenarios Actor 

Actor Intention System Responsibility 

Capture of the vehicle image System saves location of the image capture 

 System saves time of the image capture 

 

Table 4.2: Image Analysis. 

Use Case  Data Pre-processing 

Primary Actors Image Recognition API 

Pre-condition User has Internet on Platform being used 

Should be able to access the image from the camera 



 

 

38 

 

Post-condition System can display the vehicle number plate, model and 

colour. 

Main Success Scenarios Actor 

Actor Intention System Responsibility 

Perform image recognition Image is retrieved to identify the number plate, 

model and colour of the vehicle. 

 Sends the details retrieved to the system 

 

Table 4.3: Vehicle Details Classification. 

Use Case  Data Pre-processing 

Primary Actors System 

Pre-condition Receives data and stores it properly. 

 

Post-condition Display duplicate number plate, model and colour of the 

vehicle 

Main Success Scenarios Actor 



 

 

39 

 

Actor Intention System Responsibility 

Verify if data exists System checks if what has been captured has 

been captured before. 

 System stores the data if it has not been captured 

before 

 Compares the details of the capture if the same 

details are available 

 Classifies the data as duplicate or not 

4.7 System Sequence Diagram 

The figure 4.5 shows the system sequence diagram. It shows the sequence of interactions between 

the user and the proposed system as well as interactions between the various internal components 

of the system. This focuses on how the system’s processes interact to make sure that the data 

processing is achieved, and a desirable output is reached. The camera captures the image then 

passes the image to the image recognition API which in turn gives out the vehicle number plate 

model and colour. This is then sent to the system where it compares the new data to what is stored. 

If this is the first time capture the system stores the data if it’s not a comparison is done on the time 

and location. With this comparison the system can classify the number plate as a duplicate or not. 

If it is a duplicate, then the system displays it as a duplicate as the output. 



 

 

40 

 

 

Figure 4.5: Sequence Diagram. 

 

 

 

 



 

 

41 

 

Chapter 5: System Implementation and Testing 

5.1 Overview 

This chapter depicts the implementation, testing and validation of the actual prototype as proposed. 

The functionalities incorporated in the system includes the systems requirements as from the 

conceptual framework in the literature review chapter. To appreciate the feel of the system 

implementation, this section also includes screenshots of the various application which is mainly 

in the client side of the distributed system.  

The duplicate number plate detection system was built by improvising the cameras aspect and 

developing an android application that would make good use of the mobile phone camera to 

capture the images of vehicles. The system was built as described in chapter three and four. Testing 

included functional and usability tests to check if the developed system accomplished the 

objectives set out at the beginning of the research project 

5.2 Description of The Testing Environment 

The system environment is made up of both the software and the hardware. This section provides 

details on the hardware and software used and how they aid in the system performance. Describing 

the system environment will also assist in ensuring that the components are key variables in the 

system characteristics such as being scalable. Tests also performed on the model will be described 

below. 

5.2.1 Hardware Specifications 

The system runs on a test computer with a RAM (Random Access Memory) of 8 GB, this enhances 

the systems performance by enabling the API to be able to efficiently retrieve the number plate 

model and color of the vehicle with minimum time and also assisting in classification of the results 

as a duplicate or not.  For better performance, the RAM can be enhanced making the system a 

lightweight on the hardware. The hard disk space used in the test environment is a minimum of 20 

GB, this makes sure that there is enough storage for the images being captured which might grow 

in size to while trying to get more data to compare against as well as the number of vehicle’s using 

that road considering that the system will automatically delete images that have been stored for 

more than six hours. 



 

 

42 

 

A high processing speed of 2.7GHz and above is recommended and this has been made to run on 

a corei5 hardware 8th generation, this make it easier for the processors to handle different threads 

or processes that results during processing.  

For image capturing, the system uses a separate camera where by the researcher has improvised 

and created an application that can make good use of the camera. The phone camera represents the 

actual camera which captures the image of the vehicle sends them to the server for processing. For 

better performance, better fixed location cameras with higher pixels that captures clear images 

even with minimal clarity which could be caused by weather or the car being dirty. This will 

facilitate clear and accurate classification hence accurate results.   

Table 5.1: Hardware Specification. 

Hardware Minimum Requirements 

Processor Core i5 

Cycle speed 2.7GHz 

Hard Disk Space 20GB 

 

5.2.2 Software Specifications 

The proposed system runs on a windows operating system, there is no specific reason as to why 

this is preferred. Firebase real-time database has been used to maintain the client server 

architecture communication as it relays the processed data in real-time. Firebase enables clients to 

share real-time instances and automatically receive updates with the latest data.  

5.3 Prototype Development Environment 

The proposed system has a number of applications being used in order to supplement its 

productivity. The applications are built in different environments as stated below; 

(i) Android Studio - Android studio is an integrated development environment for android 

development applications. Android studio offers more features and it enhances 

productivity during development of an application. 

(ii) gRPC for the distributed system - gRPC is a modern, open source remote procedure call 

(RPC) framework that can run anywhere. It enables client and server applications to 



 

 

43 

 

communicate transparently, and makes it easier to build connected systems. It has been 

used in last mile of computing in mobile and web client since it can generate libraries for 

iOS and Android and uses standards based HTTP/2 as transport allowing it to easily 

traverse proxies and firewalls, this makes the productivity of the proposed system more 

accurate. 

5.4 Model Components 

This part describes the proposed systems components and where necessary screenshots are used 

to elaborate more on how the different components and modules of the system work, how they 

affect the systems performance and output. 

5.4.1 Image Input Components 

The proposed system requires a camera or a motion sensor to identify and capture vehicle images 

that are using a particular route. The camera should be able to send the image, time and location 

to a temporary storage before the image is sent to the API for processing. With that in mind the 

researcher developed an android application that was able to make good use of the phone 

application to capture images which is also able to send the location and time of the capture to the 

database. The use of this application also makes it easy for anyone with the knowledge of using a 

basic phone to be able to take pictures of vehicles. 

The camera will be accessed once you have installed the application in your phone and this will 

enable you to take an image of a vehicle. The image time and location is sent to the database as 

described above. The resulting android application developed was put in a test mobile device to 

test it. The application interface screen shot taken from the mobile test device is shown by the 

figures 5.1. The figure below shows the interface once you open the application. 



 

 

44 

 

 

Figure 5.1: Camera Home page. 

 

After clicking on the green button, the application will allow you to access the camera and once 

you have taken the picture it requests you to upload the image automatically. Figure 5.2 shows the 

image being uploaded. 



 

 

45 

 

 

Figure 5.2: Upload Page. 

Once the image has been uploaded successfully the application system give feedback as shown by 

the figure 5.3. 



 

 

46 

 

 

Figure 5.3: Capture Page. 

 

Figure 5.4 also shows what happens when the image captured is not a vehicle. The Camera will 

capture it because it’s a moving object but the image is rejected by the API hence no activity 

recorded. 



 

 

47 

 

 

Figure 5.4: No Caption page. 

 

5.4.2 System Components 

The system module that has been used in the proposed model is made up of the main modules that 

provides an interaction interface for the security personnel and also shows the results the data of 

the captured image which includes model, color and vehicle number plate.  The security personnel 

need to log in as shown by the figure 5.5. The main module is the client server which provides an 

interface for interaction to the environment for demonstration purposes. This module shows results 

from the API and will help in aiding the security personnel know if there is a duplicate or not and 

the personnel can decide on which action is to be taken. Figure 5.5 shows the login interface by 

the administrator or the security personnel. 



 

 

48 

 

 

Figure 5.5: System Login Page. 

 

5.5 System Testing 

The system was tested to check how the developed system performed in comparison to having the 

security personnel figuring out everything by themselves. The test was performed on the three 

modules, the mobile application that is supposed to act a camera, the system that is supposed to 

give the results as well as the API that is supposed to retrieve information from an image is 

described below. 

5.5.1 Camera Application 

The camera application was tested on the android studio and later on an actual mobile device by 

taking images and testing if they are sent to the system. The researcher also took a picture of 

images that did not have number plates and the figures below shows the results of the different 

scenarios expected when using the camera application. The images that had number plates were 

correctly labelled and images that did not have number plates were also correctly labelled.  



 

 

49 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
  Figure 5.6: Capture Page 



 

 

50 

 

 

Figure 5.7: Upload Page. 



 

 

51 

 

 

Figure 5.8: Error Page. 

 

5.5.2 Storing Captured Information 

Once the number plate is identified from the captured image the information is then stored in the 

database. The system also helps to retrieve time and location the image was taken. The Location 

is identified by the IP address of the camera that was used to take the image and hence identifying 

the location of the Image. The figure below shows the backend with a sample images of cars taken 

as well as the database. 



 

 

52 

 

 

Figure 5.9: Home page. 

Figure 5.10 show the details of the database. It shows the table structure and the relation view of 

the database. 



 

 

53 

 

 

Figure 5.10: Data base. 

5.5.2 Flagging out of Duplicate Number Plates 

The administrator or the security personnel is able to view the flagged-out vehicle image with the 

time and location of the two instances the view was captured. The user of the system must be 

logged in order to be able to view the flagged-out vehicle. The figure below shows a sample of a 

vehicle captured within the threshold time and hence it was flagged out as a duplicate vehicle.  



 

 

54 

 

 

Figure 5.11: Duplicate Number Plates. 

5.6 Performance Measure 

In model generation and building, one of the process over looked is the system validation. System 

validation is the process of checking the accuracy of something or in this case, the system. The 

researcher used functional requirements, usability requirements and supportability requirements 

to look at the validity of the system developed.  

5.6.1 Functional Requirements 

Table 5.2 shows the functional requirements for the system obtained from the validations done on 

the system. The elements assessed were based on the images captured. 

 

Table 5.2: Functional Requirements. 

Requirements Priority Results 



 

 

55 

 

Allow capture of vehicle images High 

The system was able to capture images by using 

the phone camera application which was an 

improvisation instead of using the actual 

cameras.  

System should be able to receive 

images captured, when and where. 
High 

The system was able to receive the images 

captured by the camera, the time and location the 

vehicle was captured. 

System should be able to retrieve 

the number plate, model and colour 

of the vehicle captured 

High 
The system was able to retrieve the details of the 

vehicle from the image captured 

System should be able to flag out 

vehicles with duplicate number 

plates 

High 

The system was able to flag out vehicles with 

duplicate number plates captured within the time 

threshold. 

 

5.6.2 Usability Requirements 

Table 5.3 shows the usability requirements for the system. The simplicity of use and the speed of 

use of the system 

Table 5.3: Usability Requirements. 

Requirements Priority Results 

Simplicity of use High 
The researcher designed the system to have few 

interfaces to ensure optimization of the system.  

Speed of the system Medium 
The speed of the system varied depending on 

availability of internet  

 



 

 

56 

 

5.6.3 Reliability Requirements 

Table 5.4 shows in this case the researcher wants to ascertain that the system is able to interface 

with other systems. 

Table 5.4: Reliability Requirements. 

Requirements Priority Results 

Ability of the system to interact with 

the database. 
High 

The system is able to interface properly with the 

system database via the system API.  

 

5.6.4 Supportability Requirements 

This is the ability of the system to be able run across different desktop platforms as well as 

accessing the system from different web browsers as shown by the Table 5.5. 

Table 5.5: Supportability Requirements. 

Requirements Priority Results 

Accessibility of the system from all 

the browsers 
Medium 

The system was accessible across major 

browsers; Google Chrome, Mozilla Firefox and 

Internet explorer.  

Accessibility across Major desktop 

platforms 
Medium 

The system can be deployed on Windows and 

Linux systems.  

 

 



 

 

57 

 

Chapter 6: Discussion, Conclusions and Recommendations 

6.1 Introduction 

This chapter discusses the results of the research in light of the objectives set out at the beginning. 

The researcher’s intention was to develop a system that is able to flag out duplicate number plate 

using fixed location cameras. The study was built on five objectives the core objective being to 

develop a duplicate number plate identification system using fixed location cameras in different 

locations. The researcher explains the findings from the study and goes ahead to discuss them as 

well as explaining the importance of the study.  

6.2 Evaluation of The Systems Used in Vehicle Identification 

The first objective was to evaluate the systems used in vehicle identification. In the literature 

review carried out by the researcher it was realized that there are three different systems used in 

vehicle identification the most common of them all being the use of computer vision. It is preferred 

because it is a mass observation method that makes good use of image recognition to be able to 

identify a vehicle in an image. From the findings the use of image recognition in vehicle 

identification was preferred from the others to be used in vehicle identification of the research as 

it enables use of different fe