Content uploaded by Anisa Fathi Elbokhare
Author content
All content in this area was uploaded by Anisa Fathi Elbokhare on Dec 15, 2022
Content may be subject to copyright.
Libyan Vehicle Number Plate Recognition System using (WHT)
Anisa F. Elbokhare EmadEddin Gamati Amani M. Soliman
Computer Science Department, Faculty of Education Tripoli University of Tripoli
A.elbokhare@uot.edu.ly E.gamati@uot.edu.ly a.soliman@uot.edu.ly
صلا :
( ا ت فا أ LPR ا ما )
وا . يرا،ةا LPR ا ا أ ا ءا ا
ق وا ما ت،ا و ا / ةرادإ أ،روا ضااو،ا .ذ إ و
اا ا تا ت مرأ طا ً اً
أ ةرا ا ن نأ
( تزرا ما ةرا ا ر جاا )ةر .را، ت ة ا ا
را و او ةءا را ةدو مرا لأو طأ تا يا ن
،ا ا فا إ .ا فو
ا ف آ را ح تارا ت ا ما ا تا
( فا يا فا ماOCR ) Walsh-Hadamard Transform (WHT) و .
ا تا ةاو برا ا ت ةءا ل تارا فا ح
. تاءا او ا را ةد ض ءا بأ .
Abstract:
One of the most important systems in our life today is the vehicle license plate
recognition system, which is used to identify cars based on plate numbers. This process
faces several problems due to the differences in the plate number shapes, lighting and dust,
in addition to the problem of recognizing the Arabic letters. An automatic license plate
recognition mechanism based on the Libyan system of numbering cars license plates is
proposed in this paper by using an optical character recognition (OCR) with the Walsh-
Hadamard Transform (WHT) approach. The results of the experiments were promising in
terms of identifying cars via reading (recognizing) the license plates correctly. The major
errors were related to the low quality of input images which caused wrong readings.
Keywords: license plate recognition, optical character recognition, image processing,
Walsh-Hadamard Transform.
Introduction:
License plate recognition (LPR) systems have a lot of attention from researchers
driven by the commercial demand on its applications. Recently, LPR has formed the core
part of intelligent transport systems because of its potential applications to areas such as
highway electronic toll collection, traffic management / monitoring systems, security
purposes, and so on. Operating such systems with vehicle license plate identification can
be as simple as capturing plate numbers of vehicles by a camera (generating an image) and
then extracting the plate number in the image by an image processing algorithm.
In most countries, vehicle plates consist of two metal plates that are displayed at the front
and rear of all vehicles with a unique serial number (a unique identifier number) that is
used to differentiate vehicles from each other (called vehicle plate number). The most used
pattern for numbering the plates consists of two parts (each can contain numbers and/or
letters), in addition to some limited special characters/symbols. The first part of most used
pattern, usually refers to the country/city (where the vehicle is registered); the second part
is a unique serial vehicle number (may contain letters) representing a sequence in the
registry of the Local/National Traffic Department.
Therefore, vehicle plate number recognition techniques depend on studying, analyzing and
understanding the local used shapes/patterns for vehicle plate numbering systems. Taking
in account, the parts of the numbering, the language used, special characters allowed, and
the standard different shapes of the plates.
M blihed eeach egadig he ideificai f ehicle eal lae chaace
covers English language characters. In Libya, Arabic is the Approved language for
numbering vehicle plates. Due to limitations in conducted research targeting reading
vehicle plates in Arabic, we have responded to the need to address a system for identifying
Arabic vehicle metal plate numbering in Libya.
Related work:
There are several common algorithms of vehicle license plate recognition. Sarbjit Kaur and
Skhi Ka eeed a ae A Efficie Aach f Nbe Plae Eaci
f Vehicle Iage de Iage Pceig [7]. I hi ae a aach f
preprocessed vehicle input image using morphological operations, thresholding, sobel
vertical edge detection, and connected component analysis is used for number plate
extraction. It works well under low resolution, noisy and low contrast images.
Othman Khalifa, Sheroz Khan, Rafiqul Islam and Ahmad Suleiman [5] produced a paper
Malaia Vehicle Licee Plae Recgii, he ieaial aab jal f
ifai echlg. The ah efed ecgii f licee lae de a
environmental conditions, with no assumptions about the orientation of the plate or its
distance from the camera . A simple texture-based approach based on edge information is
used for solving the problem of localization of a license plate. Then a simple multi-layer
Perceptron neural network was used to recognize the segmentation of characters.
Simulation results were shown to be an efficient method for real time plate recognition.
Birmohan Singh, Manpreet Kaur, Dalwinder Singh and Gurwinder Singh [2] published a
ae Aaic be lae ecgiion sye b chaace ii ehd. A
system for developing robust Automatic number plate recognition was suggested in this
paper. It depends on proposed a new algorithm for number plate localization which is
based on character positioning method. Character recognition is carried ou with a support
vector machine. Asyntactic analysis of number plate format for a particular geographical
region was used to solve the problem of similar shape characters.
Gaeh R. Jadha, Kailah J. Kaade [4] e Automatic Vehicle Number Plate
Recgii f Vehicle Pakig Maagee Se. The diced ig diffee
morphological operations in such a way that the number plate of a vehicle can be identified
accurately.
Safaa S. Omran, Jumana A. Jarallah, [6] preeed ae Ca Licee Plae
Recgii Uig OCR i hi ae, he ed a aaic licee lae
recognition system for Iraqi car license plates using (OCR) with a correlation approach and
templates matching for plate recognition. This was the same approach used by Sharma in
G [8] i hi ae iled Peface Aali f Vehicle Nbe Plae Recgii
Se Uig Telae Machig Techie i e f ecgii ae.
The proposed approach:
Before we dive into the proposed approach of this paper, it is worth noting that the Walsh-
Hadamard Transform (WHT) is used as an effective technique in our approach. Therefore,
we will start by introducing this technique.
Walsh-Hadamard Transform (WHT):
WHT is used in a wide variety of scientific and engineering applications. It is employed in
image processing, speech processing, filtering, and power spectrum analysis. It is very
useful for reducing bandwidth storage requirements and spread-spectrum analysis [3]. The
Walsh-Hadamard transform (WHT) is an orthogonal transformation that decomposes a
signal into a set of orthogonal, rectangular waveforms called Walsh functions. The
transformation has no multipliers and is real because the amplitude of Walsh (or
Hadamard) functions has only two values, +1 or -1. Therefore, WHT can be used in many
different applications [1] .
The Walsh-Hadamard transform (WHT) was used in this paper because it gave a better
result in Arabic character recognition [9]
Libyan Vehicle Number Plate Recognition:
The process of recognizing the metal plate starts from taking the image of the plate and
then using clarity enhancement processing techniques to get the best possible clear picture,
as most errors occurred due to the lack of clarity of the image, either due to climatic
conditions (such as dust), lighting (day/night), or even due to movement. After that, the
process of separating the contents of the image and matching it to the previously stored
database begins.
There are two standard shapes for Libyan car license plates written in Arabic, as shown in
Figures [1] and [2].
Fig. 1. The first shape of Libyan Plates
Fig. 2. The second shape of Libyan
Plates
The stages of the proposed approach:
The recognition of Vehicle Number Plate images can be divided into three main stages:
1. Preparatiojn of the database;
2. Processing the image of the Vehicle Number Plate;
3. Recognizing the contents of the Vehicle Number Plate.
The first stage: preparation of the database:
A database was created and fed with all the letters, numbers and words that are known to
be used on the Libyan Vehicle Number Plate. Some sets of the plates had complete words
(string of characters), where others have just one letter only. To give an idea of those
words and letter, here is an illustrated list:
a. Numbers from 0 to 9.
b. Dash (-)
c. Words ()
d. Characters ( )
e. Characteristics of each letter or number above, such as length and width, and WHT
spectrum coefficients [9]
The second stage is preparing the Vehicle Number Plate image includes:
1- The image was captured by the camera is converted from RGB and Gray-scale image
to a binary image
Figure 3: Converted image from RGB and Gray-scale image to a binary image
2- As Known, the Vehicle Number plates in Libya are not all in the same format. Some
plates are narrow in shape with one line and others are wide with two lines, so the
image of the plate with two lines is separated into two sub-images, each consisting of
one line.
Figure 4: Sub-images of converted two-line Libyan plate to one line
3- The data in the previous image is separated into sub-images where each contains an
individual letter or number or word. Now, the plate is ready to be Recognized.
Figure 5: Sub-images of letter, number, and word separated from plate
The third stage: image recognition
1- Select a character from the form the preparing characters in the previous stage, then
extract the character features. Find the WHT spectrum coefficients for it. [9]
2- Compare selected coefficients with stored database coefficients, if the selected
coefficients are similar to that of the reference character, Put it in the output.
3- Repeat for all next characters of image, until end of Vehicle Number Plate image.
EXPERIMENTAL RESULTS:
This method is applied to different types of Libyan Vehicle Number Plates using the
Walsh Hadamard Transform for recognition of plate characters The images for the input to
the system are colored images with variable sizes. The test images were taken under
various conditions obtained from different distances from the camera and with different
angles and lighting ratios.
The database used to test the system was 75 images; 68 images were obtained
correctly with 90.67% success. The major errors were related to bad quality input images
which caused wrong input in the second stage.
(a) (b)
Figure 6: Experimental Results (a) Input plate (b) System output
Conclusions:
Although there are many systems for recognizing various plates all over the world
(country-based designs), the proposed effort is directed to Libyan license plates. The
license plate recognition involves image acquisition, license plate extraction, segmentation,
and recognition phases. Beside the use of Arabic, Libyan license plates have several
unique features that had to be taken care of in the recognition phases.
In this paper, a system of Vehicle Number Plate Image Recognition was proposed.
Preparation captured plate image, then the data was separated into sub-images where each
one contained a letter, number or word WHT spectrum. coefficients were compared with
stored database coefficients. If the selected coefficients were similar to that of the reference
character, the system stored the character. This was repeated for all the next characters of
image until Vehicle Number Plate image is completed. The proposed approach was tested
over a large number of car number plates and the ratio of success was 90.67%. Bad input
images caused most errors.
References:
1. A.A.C.A. Jayathilake 1, A.A.I. Perera 2, M.A.P. Chamikara, Dicee Walh-
Hadaad Taf i Sigal Pceig, IJRIT Ieaial Jal f Reeach
in Information Technology, Volume 1, Issue 1, pp. 80-89, January 2013,
2. Birmohan Singh, Manpreet Kaur, Dalwinder Singh and Gurwinder Singh, Aaic
number plate recognition system by character position method, I. J. Computational
Vision and Robotics, Vol. 6, Nos. 1/2, 2016.
3. Eael Aig Fig, Phili Ya Okee, Jh Aad, Walh-
Hadamard Transform and Support Vector Machine Based Rotor Angle Stability Status
Prediction, Ieaial Research Journal of Engineering and Technology (IRJET),
Volume: 5, Issue: 7, July 2018.
4. Gaeh R. Jadha, Kailah J. Kaade, Aaic Vehicle Nbe Plae Recognition
for Vehicle Pakig Maagee Se, Ce Egieeig ad Iellige
Systems ISSN 2222-1719 (Paper) ISSN 2222-2863 (Online), Vol.5, No.11, 2014.
5. Othman Khalifa, Sheroz Khan, Rafiqul Islam and Ahmad Suleiman, Malaia
Vehicle License Plae Recgii, The international arab journal of information
technology, Vol.4, No. 4, October 2007.
6. Safaa S. Omran, Jumana A. Jarallah, Iai Car License Plate Recognition Using
OCR, Ciha Uiei-Erbil Scientific Journal, Special Issue No.1, July 2017.
7. Sarbjit Kaur, Sukhvir Kaur, A Efficie Aach f Nbe Plae Eaci f
Vehicle Iage de Iage Pceig, Ieaial Jal f Ce Sciece
and Information Technologies, Vol. 5 (3), 2014, 2954-2959.
8. Shaa G, Pefance Analysis of Vehicle Number Plate Recognition System
Uig Telae Machig Techie, J If Tech Sfae Eg 8: 232, )2018).
doi:10.4172/2165-7866.1000232.
9. Wajdi S. Besbas, Anisa F. Elbokhare, Mohamed R. Sunni, EmadEddin A. Gamati,
Performance Comparison of WHT and DCT Transforms Used in Sliding Window
Pied Aabic OCR, Ieaial Sciece ad Techlg Jal, l 10, Ail
2017.