Bhupendra Moharil et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol.

3 (5) , 2012,5108 - 5115

An Efficient Approach for Automatic Number

Plate Recognition System Using Quick Response
Codes
Bhupendra Moharil, Vijayendra Ghadge, Chaitanya Gokhale, Pranav Tambvekar
Department of Computer Engineering, M.I.T.C.O.E., University of Pune, Pune, Maharashtra, India

Abstract— The ANPR (Automatic Number Plate Recognition were let to produce industrial systems, first at EMI
System) can be called as a car screening technique that Electronics, and then at Computer Recognition Systems
uses optical character recognition(OCR) on images to (CRS) in Workingham, UK. Early trial systems were
read vehicle registration plates. The volume of automobile deployed on the A1 road and at the Dartford Tunnel. [1]
crimes has been on the increase and many new crime detection
techniques have evolved on its account. High-speed car chases,
automobile thefts are now-a-days very common. The best way B. ANPR System using OCR
to keep track of the cars is their registration numbers. The At the hub of the system is the OCR (Optical Character
number plate recognition system is the most suitable solution Recognition system) which is used to extract the
for such cases. alphanumeric characters present on the number plate. To do
The number plates often get corroded and the detection this it first uses a series of image manipulation techniques
process becomes erroneous. Our contribution towards ANPR to detect, normalize and enhance the image of the number
is the inclusion of the QR codes (Quick Response codes) on the plate. There are two components in the system, the cameras
automobiles, fine-tuning the detection technique. The at the front-end and the remote computers at the back-end.
proposed approach provides quick response, overcomes the
noise issues in image processing and at the same time fine
Usually two cameras are used at a time to increase
tunes the detection technique. efficiency. The cameras as shown in the Fig. 1 just perform
Keywords— ANPR, OCR, QR codes, CCTV’s the task of capturing the images of number plates and
sending it to the remote computers. The remote computers
I. INTRODUCTION then perform further operations like OCR on the stored
ANPR stands for Automatic number plate recognition images sent by the cameras at the lane-level. In order to
system. License Plate Recognition (LPR) is an important process the high amount of images stored, a “server farm”
function in intelligent traffic control systems. These is used which comprises of many computers working
systems have many applications such as parking together. An example of a server farm can be the London
management systems, access control, border control and Congestion Charge project. The remote computers can be
monitoring, and tracking vehicles. ANPR is also known as: linked with the database which stores the details of the car
1. Automatic vehicle identification (AVI) owners and thus the required information can be obtained.
2. Car plate recognition (CPR) Using this information the culprit can be caught. [1], [11]
3. License-plate recognition (LPR)
4. Automatic license-plate recognition (ALPR)
5. Lecture Automatique de Plaques
d'Immatriculation (LAPI)

ANPR Systems are the most widely used systems all

across the world. A typical number plate detection system
involves the use of OCR technique. With OCR techniques
in operation, its performance is governed by the image
quality. In turn the environmental condition affecting the
number plate that directly hamper the image taken by the
camera entirely governs the efficiency of the detection
approach. With addition of QR codes, we propose a novel Fig. 1 CCTV cameras
method where the problems arising in the detection are
reduced to a great extent. The paper discusses the current 1. The existing ANPR system consists of two major
OCR technique and the number recognition systems that are blocks as shown in the Fig. 2:
in use at present with the applications. The experimentation a. Lane level
done with our approach has shown results that are worth to b. Remote computers
notice. 2. At the lane level, the CCTV (Closed Circuit Television)
cameras take a snapshot/video feed [16], [19] of the car
II. EXISTING METHODOLOGY in such a way that the number plate is also included in
A. History the image. A minimum of two cameras, each fitted
ANPR finds its origin in UK. It was first developed by with infrared illuminators so they can work day or
the Police Scientific Development Branch in 1976. night, are mounted on gantries above the road, either at
Prototype systems were working by 1979, and contracts

5108
Bhupendra Moharil et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 3 (5) , 2012,5108 - 5115

the roadside or in the central reservation. Speed 3. The optical recognition technique is performed at the
cameras work out the vehicle's average speed, given remote computer level and consists of the following
the time it takes to drive between the two camera algorithms: [15]
positions. As vehicles pass between the entry and exit a. Plate localization: Isolates the number plate from
camera points their number plates are digitally the snapshot. [18]
recorded, whether speeding or not. Then, by ANPR b. Plate orientation and sizing: The isolated number
recognition, the images on the video of matching plate is then resized and rotated as per the
number plates are paired up, and because each image requirements of the recognition software.
carries a date and time stamp, the computer can work c. Character Segmentation: Each character on the
out the cars average speed between the cameras.[9], [17] number plate is separated by drawing vertical and
horizontal edges and thus forming rectangles
around the character as shown in the Fig. 3.
Search for number plate

LANE LEVEL
Fig. 3 Recognition of the number plate using OCR
Snapshot/Video feed d. Normalization: After character segmentation the
brightness and contrast of the image is adjusted.
e. Character Recognition: Each character is then
recognized by the software. The software
implements the technique of pattern matching to
Image/video sent to detect the characters, thus matching the characters
servers with the stored samples. The characters are then
combined together to form the number plate [20].
f. Syntactical analysis: Checks characters and
positions against country-specific rules.

Plate localization 4. The percentage of recognition is as follows [4]:

TABLE 1
PERCENTAGE OF ANGULAR RECOGNITION USING OCR [4]

Plate orientation and Angles Plate Number Plate

sizing (Degree) Localization (%) Recognition (%)
5.00 96.00 95.65
7.50 96.00 91.30
Character Segmentation 10.00 96.00 86.95
12.5 96.00 82.61
15 96.00 78.26
OCR AT
From the table 1 we infer that the efficiency of the
Normalization system reduces as the angle goes on increasing.

C. Applications of OCR
REMOTE
The ANPR system using OCR has the following
Character Recognition applications [28]:
1) Crime investigation:
The ANPR can be used for tracking of lost vehicles. In
case of a car/vehicle theft, the car can be detected by
COMPUTER LEVEL searching in the vicinity of the area by using ANPR system.
This can help in speeding up the process of searching thus
Syntactical/Geometrical reducing the workload of the police force.
analysis 2) Illegal racing:
The practice of illegal car sale in the black market is not
Fig. 2 Overview of the automatic plate recognition unheard of. The cars are dismantled and exported part by
technique [4] part to other countries and then such cars are used for illegal
racing and other money-making / entertainment purposes.

5109
Bhupendra Moharil et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 3 (5) , 2012,5108 - 5115

ANPR provides a strong mechanism to put a stop to these is considered in edge analysis. Character segmentation is an
activities. important step in license plate recognition (LPR) system.
3) Over-speeding: [14]
If a police officer detects an over-speeding car he can Retrieving the literature it was evident that, the previous
use ANPR system to track down the car details and fine the system has some kind of limitation like it is plate size
driver. This can help keep the accidents at bay. dependent, colour dependent, works only in certain
4) Parking in the NO PARKING AREA: conditions or environment like indoor images etc.
The cars parked in the no parking zones can be detected The ANPR system requires a large database to store the
even though no ground force is available on site. information. The National ANPR Data Centre uses an
5) Identification of traffic density : Oracle database, with data for analysis extracted to a
ANPR system can also be used to identify the low and high PostgreSQL database to avoid affecting core performance,
traffic density areas, based on the number of vehicles and custom-written Java software with specific police
passing through a particular CCTV camera coverage per networks rather than the internet used to send and receive
unit time. Thus if the number of plates detected by a camera data [12].
in unit time is greater than usual then it implies high traffic The best example of automatic number plate detection
density otherwise the density may be low. system can be the London Congestion Charge Project.
6) Electronic toll collection: Transport for London (TfL) uses ANPR systems and
ANPR system is also employed for the electronic toll charges motorists a daily fee of £10 paid before 10pm if
collection system. This reduces the manual labor increasing they enter, leave or move around within the congestion
the efficiency of the toll collection system. An example can charge zone between 7 a.m. and 6:00 p.m., Monday to
be the London Congestion Charge Project. Friday. A reduced fee of £9 is paid by vehicle owners who
sign up for the automatic deduction scheme. Fines for
III. LITERATURE SURVEY traveling within the zone without paying the charge are £60
Since the mid 90’s a lot of research works has been per infraction if paid before the deadline, doubling to £120
carried out in the field of Vehicle Number Plate (VNP) per infraction thereafter [7].There are currently 1,500
recognition. The manufacturers of SPECS CCTV cameras cameras, which use Automatic Number Plate Recognition
quote that the SPECS system is so efficient and user (ANPR) technology. There are also a number of mobile
friendly that Manchester CTO processed 2,500 offences in camera units which may be deployed anywhere in the zone.
6 man hours [9]. Based on the statistics describing the use of ANPR system
Image Acquisition is the first step in an ANPR system [8], an example is discussed. A scenario where a car is
and there are a number of ways to acquire images. The parked in a no-parking area and a CCTV camera snaps a
existing literature discusses different image acquisition picture of its number plate at 7:34 p.m. Then within 250
methods. Yan et. al. [21] uses an image acquisition card milliseconds the image is sent to the remote computer along
that converts video signals to digital images based on some with the date and time. 1 second of conventional OCR
hardware-based image pre-processing. Naito et. al. [22], processing generates 18 seconds of editing time (7.50+10.5)
[23], [24] developed a sensing system, which uses two and 3.4 errors (8.4-5.0) that get past manual error correction.
CCDs (Charge Coupled Devices) and a prism to split an After OCR the license number is obtained. The Police
incident ray into two lights with different intensities. The officer in charge may then track the exact owner of the car
main feature of this sensing system is that it covers wide by accessing the recognized number plate record and thus
illumination conditions from twilight to noon under take a strict action against him. This can be done in a matter
sunshine, and this system is capable of capturing images of of seconds based on the server workload and network
fast moving vehicles without blurring. Salgado et. al. [25] congestion.
uses a Sensor subsystem having a high resolution CCD The ANPR system using OCR was found to have the
camera supplemented with a number of new digital following shortcomings /disadvantages:
operation capabilities. However the installation of such a 1) Misidentification
device is costly. • In case the number is read partially, the
Kim et. al. [26] uses a video camera to acquire the image. remote computer might identify the number
Comelli et. al. [27] uses a TV camera and a frame grabber plate incorrectly or would not be able to
card to acquire the image for the developed vehicle LPR decrypt at all.
system [15]. But processing on the video feed is more
complex as every frame needs to be analysed and best
frame is to be chosen for analysis. If a wrong frame is
chosen then the recognition becomes erroneous.
For extracting the number plate from the detected image,
the LPL technique is used. License Plate Localization (LPL)
methods are broadly classified into Morphology based LPL
methods, Edge statistics, and neural networks and fuzzy
based, template based and so on. The aspect ratio, colour,
variance, edge density are some of the license plate features
used by these methods. High contrast between characters
and background in a license plate is a strong feature which Fig. 4 Misidentification

5110
Bhupendra Moharil et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 3 (5) , 2012,5108 - 5115

• Characters and numbers may be detected Hence there is a need for development of a new
wrongly and interchanged identity may occur; technique that would be capable of masking the drawbacks
characters may be read as numbers and vice of the existing ANPR system.
versa. The fig. 4 shows a sample that
describes the example. E.g.: ‘o’ as ‘0’, ‘5’ as IV. PROPOSED METHODOLOGY
‘S’ Taking into consideration, the shortcomings of the
2) Hazy images: existing ANPR system we propose the methodology of
Hazy images can also make the detection process using QR code for the recognition of vehicles and their
erroneous or there is a possibility of no detection at all. An respective owners.
example for such cases is the Fig. 5.
A. QR codes (Quick response codes):
1. QR code is trademark of matrix type (two-
dimensional) barcode. [13]
2. More recently, the system has become popular
outside the industry due to its fast readability and
large storage capacity compared to standard UPC
barcodes. The code consists of black modules
(square dots) arranged in a square pattern on a
white background. The information encoded can
be made up of four standardized kinds ("modes")
of data (numeric, alphanumeric, byte/binary), or
Fig. 5 Hazy image through supported extensions, virtually any kind of
3) Flaws in angular detection: data.
3. Invented in Japan by the Toyota subsidiary Denso
Wave in 1994 to track vehicles during the
manufacturing process, the QR Code is one of the
most popular types of two-dimensional
barcodes.[2]
4. It was designed to allow its contents to be
decoded at high speed.

Fig. 6 Angular detection not possible

Angular detection is not possible in case of ANPR as

the rectangulation algorithm, implemented in OCR is not
possible thus characters may be misread/ overlapped. The
character recognition fails even if the angle made by the
camera’s line of sight with the vertical is a mere 150 angle.
Thus the possibility of catching a wrong person increases
substantially.
E.g. ‘T’ may be read as ‘I’ or ‘C’ as ‘O’ or ‘2’ as ‘1’ or
‘G’ as ‘C’ or ‘C’ as ‘O’ as shown in the Fig. 6.
4) Problems in recognition under bad weather
conditions:
Natural Calamities like heavy rainfall or snowfall may Fig. 7 Sample QR code embedded with information
hinder the detection system. “Paper by: Pranav, Chaitanya, Vijayendra, Bhupendra”
5) High speed cameras are required hence general cost
of deployment is high. Unlike the old bar code that was designed to be
6) Optical character recognition (OCR) is slow. mechanically scanned by a narrow beam of light, the QR
7) Depends highly on the efficiency of camera. code is detected as a 2-dimensional digital image by a
E.g.: semiconductor image sensor and is then digitally analyzed
• Dust settled on the camera lens. by a programmed processor. The processor locates the three
• Power cuts/battery discharged distinctive squares at the corners of the image, and
• Short circuits normalizes image size, orientation, and angle of viewing,
• Low resolution with the aid of a smaller square near the fourth corner. The
8) Non-standardized number plates. E.g. Number small dots are then converted to binary numbers and
plates written in different scripts like ‘Devnagari’ validity checked with an error-correcting code. A sample
cannot be recognized by the system. QR code can be fig. 7.
9) Duplication of number plates is easily possible.

5111
Bhupendra Moharil et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 3 (5) , 2012,5108 - 5115

The amount of data that can be stored in the QR Code 1. Faster scanning.
symbol depends on the data type (mode, or input character 2. Quicker response of the system due to faster
set). recognition.
The data types that can be stored on the QR codes are 3. Angular Detection possible
numeric, alphanumeric and binary. 4. Detection is possible even if the QR code is
damaged, distorted as shown in fig. 10 or mud
B. How do QR codes work? smeared.
The format information records two things: the error
correction level and the mask pattern used for the symbol.
Masking is used to break up patterns in the data area that
might confuse a scanner, such as large blank areas or
misleading features that look like the locator marks. The
mask patterns are defined on a 6×6 grid that is repeated as
necessary to cover the whole symbol as shown in fig. 8.
Modules corresponding to the dark areas of the mask are
inverted. The format information is protected from errors
with a BCH (Bose and Ray- Chaudhuri code, a code which
is a cyclic error correcting code) [5], and two complete Fig. 10 Damaged QR code
copies are included in each QR symbol. 5. Scanning possible even if the camera resolution is
less which may not be possible in case of OCR
technique.
6. Each QR code is unique, due to which there is
least possibility for misidentification.

D. Proposed approach towards inclusion of QR codes in

ANPR:

Fig. 8 Format information

The message data is placed from right to left in a zigzag

pattern, as shown in fig. 9. In larger symbols, this is
complicated by the presence of the alignment patterns and
the use of multiple interleaved error-correction blocks.

Fig. 11 Block diagram for QR code detection

1. The ANPR system using QR code will consist of

Fig. 9 Message placement within a QR symbol two major blocks as shown in the Fig. 11:
a. Lane level
Four-bit indicators are used to select the encoding mode b. Software Level
and convey other information. Encoding modes can be 2. At the lane level, the CCTV (Closed Circuit
mixed as needed within a QR symbol. Television) cameras take a snapshot of the car in
such a way that the QR code is also included in the
C. Advantages of QR codes over OCR image. Otherwise a small optical QR code scanner
The existing ANPR system may fail to identify the can be installed for on spot recognition. As
vehicle under unsuitable circumstances as mentioned in vehicles pass under the camera/scanner their QR
the Literature Survey. The advantages of QR codes over codes are digitally recorded.
OCR are as follows:

5112
Bhupendra Moharil et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 3 (5) , 2012,5108 - 5115

3. The image/video/QR code embedded information 5. Logos can also be included on the QR codes as in
is sent along with the date and time stamp to the fig. 14.
remote computer.
4. The QR code recognition technique is performed
at the software level and consists of the following
steps:
5. If direct information is sent from lane level
forward it for searching the record in database.
6. Else if image/video feed is received then focus the
QR code recognition software component on to the
QR code received to retrieve the information and
then search for the record in database.

E. Proposed implementation of QR codes in ANPR system

Fig. 14 Logos included on the QR code
1. The QR codes can be embedded on the back side
beside the company logo for every vehicle.
6. Q-R codes can be printed using radioactive
2. The QR codes can also be printed on the number
material e.g. Radium to improve detection even in
plate itself or on inner side of the windshield.
unfavourable weather conditions (Rainfall,
3. Q-R codes can be painted using invisible paint on
snowfall). Radioactive ink is made by mixing a
the roof (as shown in fig. 12) of the vehicles
small amount of 32P with waterproof black
during the manufacturing process, which can be
drawing ink. Based on the wavelength of the ink it
detected under specific wavelength of light. This
can be detected by the cameras. Thus, even if dust
can help in overhead detection.
particles are settled or mud smeared on the vehicle
the QR code can be detected. Using this technique,
night time detection is also possible.

F. Special Applications of Automatic Number Plate

Recognition system using QR codes(Dynamic Scanning):
Dynamic scanning is searching of a particular QR code
among all the QR codes passing by. If a small detection
system is installed at prime locations then a stolen car or
suspicious car can be tracked at further locations by
continuously scanning of the suspicious QR code. If
detected, an alarm will activate at the RTO office and
inform them about the location of the car.
This system when clubbed along with the GPS can help
in catching the culprit faster.
Police patrol vehicles can be equipped with QR code
Fig. 12 Invisible paint detection scanners for dynamic scanning.

4. QR codes can also be printed in stylish formats on G. Tackling the problem of Number plate duplication
the car.
E.g. Q-R codes can be printed as vinyl as shown in the As mentioned in the Literature Survey, duplication of
fig. 13. number plates is easily possible. Similarly QR code
duplication is also possible. This can become a threat
for the proposed methodology. Hence we suggest the
algorithm for avoiding the same.

Algorithm to avoid duplication:

1. Detect the QR code on the vehicle.

2. Scan the QR code and retrieve the embedded
information.
3. Save the scanned information in the database along
with the time stamp.
4. The QR codes detected would be duplicate if the
same QR code is scanned more than once within a
particular time span.
5. For this the region in which the car can be detected
Fig. 13 QR code embedded in stylish vinyl during the time span needs to be fixed otherwise

5113
Bhupendra Moharil et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 3 (5) , 2012,5108 - 5115

the same car could be detected and reported as a 3. Otherwise a unique code can be stored on the QR
duplicate. code and then the RTO officers can identify the car
6. If the same QR code is detected beyond the limits owner. The sample of such QR code can be Fig. 17.
of the calculated region then it would imply a
duplicate QR code.
7. In order to define the maximum possible distance
that the car can travel in a particular time span, we
need to calculate distance based on its maximum
speed.
8. A circle of radius equal to the maximum distance
possible is then plotted on the map surrounding the
tracked location as shown in Fig 15.
9. Now to detect a duplicate QR code we search the
region (country/state) excluding the plotted
circular region.
10. If a match is found then the code can be checked
for validity and the criminal can be penalized. Fig.17 Vehicle ID: 123454567890 [6]

4. The QR codes can be scanned even by using low

resolution camera. Thus anyone can scan the QR
code through his mobile phone/camera and
immediately lodge a complaint in cases such as:
1) Vehicle parked in a private area.
2) Unidentified vehicle found.
3) Rash driving and other malpractices
This will help the authorities in charge to pin-
point the location of the vehicles reported. A
Block diagram of the same is shown in the
fig. 18.

Fig. 15 Algorithmic view for finding of duplicates

H. Methods for information embedding in QR codes

1. The QR codes may contain URL. The URL will be
a link to a particular page on the remote server
through which the necessary information can be
obtained.
2. Another method could be directly putting the
information on the QR codes as shown in the Fig.
16.

Fig. 18 capturing the image through cell phone

V. EXPERIMENTAL RESULTS
To support our proposed approach we carried out
experimentation on the angular detection of QR codes.
As mentioned in the Literature Survey, the maximum
range of accurate number plate detection using OCR is -
Fig.16 Vehicle No.MH-12 AB 1234 Name: Tom R. 150 to 150 i.e. total 300. But with the inclusion of the QR
Licence No.123456789 Expiry Date: 1 Jan 2013 codes the maximum range becomes -37.420 to 37.420 i.e.
Contact No.9876543210 Vehicle registration 74.840. Also the results show 100 % detection and
No.ABCD1234 [6] beyond 74.840 there is no detection possible at all. Thus
the chances of incorrect detection are completely
eliminated.

5114
Bhupendra Moharil et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 3 (5) , 2012,5108 - 5115

TABLE 2 [12] Database used for ANPR system:

http://www.guardian.co.uk/technology/2007/feb/15/epublic.guardia
EXPERIMENTAL RESULTS OF ANGULAR QR CODE nweeklytechnologysection
DETECTION [13] Lakshmi Chetana Vemuri, "Geometric Invariant Digital Image
Angles Number of QR code Watermarking Techniques for QR code" (IJCSIT) International
(Degree) Samples Recognition (%) Journal of Computer Science and Information Technologies, Vol. 3
(1) , 2012,
10.00 10 100 3037 - 3041 ISSN:0975-9646
20.00 10 100 [14] Anish Lazrus, Siddharth Chaubey (IJCSIT) International Journal of
30.00 10 100 Computer Science and Information Technologies, Vol. 2 (4) , 2011,
1494-1497 "A Robust Method of License Plate Recognition using
37.4 15 100 ANN" ISSN: 0975-9646
50.00 3 0 [15] Kuldeepak, Monika Kaushik, Munish Vashishath,"License Plate
Recognition System based on Image Processing Using Labview"
The experimentation is done using a 3.2 MP resolution International Journal of Electronics Communication and Computer
Technology (IJECCT) Volume 2 Issue 4 (July 2012)
camera android phone using the application “Barcode [16] Christos-Nikolaos E. Anagnostopoulos, Ioannis E. Anagnostopoulos,
Scanner v4.1” [10]. The results for the same are as Ioannis D. Psoroulas, Vassili Loumos and Eleftherios Kayafas
described in the table 2. "License Plate Recognition From Still Images and Video Sequences:
A Survey" IEEE transactions on intelligent transportation systems,
VOL. 9, NO. 3, September 2008
VI. CONCLUSION [17] C. A. Rahman, W. Badawy, A. Radmanesh, ”A real time vehicle's
The paper highlights the drawbacks of the conventional license plate recognition system,” Proceedings. IEEE Conference on
ANPR system using OCR technique and proposes an Advanced Video and Signal Based Surveillance, Miami, Florida
efficient method with inclusion of QR codes. 2003.
[18] Samiksha Soni,Mohammad Imroze Khan", Content Based Retrieval
Through the experiments we carried out on the angular for
detection, accuracy of 100% is achieved. Even the response Number Plate Extraction of Vehicle", International journals of
time is low and the issues of image processing are resolved. Advances in Electronics Engineering.
With addition of invisible ink or radioactive ink night time [19] I-Chen Tsai, Jui-Chen Wu, Jun-Wei Hsieh, and Yung-Sheng Chen
"Recognition of Vehicle License Plates from a Video Sequence",
detection and detection in extreme climatic conditions can IAENG International Journal of Computer Science, 36:1,
be made possible. IJCS_36_1_04
Considering the positive aspects of the QR, the proposed [20] Hadi Sharifi Kolour,Asadollah Shahbahrami "An Evaluation of
method if brought into actual practise will definitely prove License Plate Recognition Algorithms", International Journal of
Digital Information and Wireless Communications (IJDIWC) 1(1):
to be a boon. 247-253 The Society of Digital Information and Wireless
Communications, 2011(ISSN 2225-658X)
ACKNOWLEDGEMENT [21] Yan, Dai, Hongqing, Ma., Jilin, Liu, and Langang, Li, "A high
We thank Prof. Prachi Joshi, Computer department, performance license plate recognition system based on the web
technique, Proceedings IEEE Intelligent Transport Systems, pp.
M.I.T. College of Engineering, Pune for her guidance and 325-329, 2001.
Prof. Kranti Dive, Computer department, M.I.T. College of [22] Naito, T., Tsukada, T., Yamada, K., Kozuka, K., and Yamamoto, S.,
Engineering, Pune for her support. "License plate recognition method for inclined plates outdoors",
Proceedings International Conference on Information Intelligence
and Systems, pp. 304-312, 1999.
REFERENCES [23] Naito, T. Tsukada, T. Yamada, K. Kozuka, K. and Yamamoto, S.,
[1] Shriram Kishanrao Waghmare, A. K. Gulve,Vikas N.Nirgude "Robust recognition methods for inclined license plates under
“Automatic Number Plate Recognition (ANPR) System for Indian various illumination conditions outdoors Proceedings
conditions using support vector machine (SVM)” [ISSN 2250 - IEEE/IEEJ/JSAI International Conference on Intelligent Transport
3765] International Journal of Computer Science and its Systems, pp. 697-702, 1999.
Applications [24] Naito, T., Tsukada, T., Yamada, K.s Kozuka, K., and Yamamoto, S.,
[2] QR codes: http://en.wikipedia.org/wiki/QR_code "Robust license-plate recognition method for passing vehicles under
[3] OCR: http://en.wikipedia.org/wiki/Optical_character_recognition outside environment", IEEE Transactions on Vehicular Technology,
[4] Journal of the Eastern Asia Society for Transportation Studies, vol: 49 Issue: 6, pp: 2309-2319, 2000.
Vol.8, 2009 “Travel Time Measurement in Real-Time using [25] Salagado, L., Menendez, J. M., Rendon, E., and Garcia, N.,
Automatic Number Plate Recognition for Malaysian Environment”. "Automatic car plate detection and recognition through intelligent
[5] Bose, R. C.; Ray- Chaudhuri, D. K. (March 1960), "On A Class of vision engineering", Proceedings of IEEE 33r Annual International
Error Correcting Binary Group Codes", Information and Carnahan Conference on Security Technology, pp. 71-76, 1999.
Control 3 (1): 68–79, ISSN 0890-5401 [26] Kim, K. K., Kim, K. I., Kim, J.B., and Kim, H. J., "Learning based
[6] QR code generator: www.QRu.nu approach for license plate recognition", Proceedings of IEEE
[7] London Congestion Charge Project, England: Processing Society Workshop on Neural Networks for Signal
http://www.roadtraffic-technology.com/projects/congestion/ Processing, vol. 2, pp: 614-623, 2000
[8] OCR : http://primerecognition.com/augprime/cost_justification.htm [27] Comelli, P., Ferragina, P., Granieri. M. N., and Stabile, F., "Optical
[9] http://www.speedcamerasuk.com/SPECS.htm recognition of motor vehicle license plates", IEEE Transactions on
[10] Scanner software(Android): http://code.google.com/p/zxing Vehicular Technology, vol. 44, no. 4, pp: 790-799, 1995.
[11] Prathamesh Kulkarni (Student Member, IEEE), Ashish Khatri, [28] http://en.wikipedia.org/wiki/Automatic_number_plate_recognition
Prateek Banga, Kushal Shah,” Automatic Number Plate
Recognition (ANPR) System for Indian conditions”, IEEE
Transactions, 2009.

5115