Automated Number Plate Recognition System

Ananya Kotibhaskar, Durvank Belitkar

Computer Science & Amogh Waykar Computer Science &
Aryan Banafal
Engg. Deptt Engg. Deptt
Computer Science & Computer Science &
Maharashtra Institute of Maharashtra Institute of
Engg. Deptt Engg. Deptt
Technology ,Pune Technology ,Pune
Maharashtra Institute of Maharashtra Institute of
Kotibhaskar.ananya@g Technology ,Pune Technology ,Pune
mail.com
[email protected]
om

Abstract— It might be challenging to locate speeding drivers number for several states, in Europe and other continents. Our
and those who have breached the law. Traffic signals and car project main objectives are to acquire in a successful way car
owner verification are now major issues in every country. As a images, extract the license number plate from the acquired
result of the speed of cars, it is difficult for police to collect image, and recognize the license plate characters by matching
numbers from moving cars, so finding methods to detect such
them with prerequired information.
people is fruitless. To address this issue, it is crucial to advertise
Automatic Number Plate Recognition (ANPR). It might be
challenging to locate speeding drivers and those who have II. Literature Survey
breached the law. Traffic signals and car owner verification are
now major issues in every country. The detection and recognition [3] S.Roy, A. Choudhury, J. Mukherjee. The proposed a
of a vehicle License Plate (LP) is a key technique in most of the system to localization of number plate and segmented the
applications related to vehicle movement. Moreover, it is a quite numbers as to identify each number separately
popular and active research topic in the field of image
This paper presents an approach based on simple and
processing. Different methods, techniques and algorithms have
been developed to detect and recognize License Plates. Vehicle efficient morphological operation and sobel edge detection
License Plate Recognition (VLPR) process is a challenging method.He also presents a simple approach to segmented all
process because of variations in viewpoint, shape, colour, the letters and numbers used in the number plate. After
multiple formats and non-uniform illumination conditions at the reducing noise from the input image we try to enhance the
time of image acquisition. contrast of the binarized image using histogram equalization.

I. Introduction [7] D. Jiang,T. M. Mekonnen,T. E. Merkebu,A

[14]Vehicles are widely used in all areas of production as well Gebrehiwot. Discussed paper presents about car plate
as in our daily lives. Therefore, the vehicle License Plate (LP) recognition system. it describes, design algorithm and future
number is an efficient way to identify vehicles, which is of implementation.
unique for each vehicle. With the rapidly increasing number The system has colour image inputs of a car and the output
of vehicles, traffic violations appear more frequently in public has the registration number of that car.
traffic, such as fraud tolls in highways or parking, speeding, The three main steps to achieve this are :
theft of cars, etc. Therefore, it is necessary to identify the Plate localization
License Plate of vehicles for safety. In everyday life, Character Segmentation
automatic license plate recognition is useful. Law enforcement, Character Recognition
traffic management, limited area access, e-tolls, and parking
lot check-in are just a few of the uses that depend on being Lippaiová et al developed a desktop application to identify
able to identify a vehicle or its owner. A recent development the license plates of automobiles that enter from the entrance.
in automated transport systems is ALPR. [1] The system During the different lighting conditions, vehicles in the video
employs image processing to identify a car from a license were captured on the exterior. For extracting the license plate
plate image , but there are numerous other variables, including of the vehicle, the system needs some time
the location, intensity, light, type, size, and colour of the
license plate, as well as environmental variations. The aim of
this paper is to develop an algorithm for license plate
recognition, particularly for an auto service. This is achieved
using an automatic number plate recognition library easyocr
and imutils. This is a system that implements the registration
 bus, scooter or motorbike. In, Soble filter technique is useful
in recognizing the type of vehicle accurately. Edges of a
vehicle can be found by this technique. There are some
techniques that are used to find out the model of the vehicles
such as -The Contour let Transform and Support Vector
Machine. To get fully assured about these techniques, these
were practically done & analyzed. In, Maximum Average
Correlation Height filter and Log r-theta Mapping methods
were applied to analyze the category of automobiles. For
revealing of region of interest in messy/jumbled situation,
MACH filter was used.
[4] Ch. Sekhar et al developed a software application to
improve security system. If an intruded vehicle enters the [2] In, Optical Character Recognition method, which is a
system alerts the security services and sends an email within widely used tool for mechanical or electronic conversion of
few times. The system can be deployed to the small scale as images of typed, handwritten or printed text into machine-
well as large scale areas. The system also ensures that the encoded text, whether from a scanned document, a photo of a
vehicle is authorized or not. The information of authorized document, a scene-photo or from subtitle text superimposed
vehicle is stored in the database. on an image. OCR software pre-processes the images to
enhance the chances of successful recognition. The two non-
[13] Puranik et al introduced a system to keep follow up of intersecting images data sets were used to copy the actual-
number and the kind of vehicles in the premises with their world cases where the neural network will be subjected to.
owners. It keeps the track of the exact time owners/drivers had Artificial Neural Networks are vastly used intelligent
left premises. This is useful if the vehicles are stolen. The calculating design for recognition of patterns. The best
system works as: the primary step is localization of number common used ANN is the multilayer feed-forward neural
plate region and character segmentation. The second step is network which has a meek structure that can categorize inputs
comparison with templates during which some templates are into a set of target groupings. Obtained license plates are split
stored in database. Third step is to know it is an exact match into individual characters using a region-based method. The
or not. In the final step vehicle number plate is detected. detection method uses a combination of template matching
algorithms and adaptive iterative thresholding.

Basically, the workings completed in and use information

mining to manage the contributions of neural network
separately, the prior one is the supreme basic method used for
neural network, which can attain decent enactment even under
rough situations. In, recognition rate can be improved by a
two stages hybrid OCR scheme. This scheme includes the
independent recognition of input character by four statistical
sub classifiers and then uses the Bayes’ method to combine
the results. Moreover, if the documented character from the
first step fit to the collections of similar characters (e.g. A/4,
B/8 and S/5), then a operational stage is used for a further
differentiation.

When used on images with poor quality backgrounds,

current techniques are ineffective. low accuracy It takes a long
time to manually evaluate something. There was a lot of noise.
High contrast photos are not produced by these techniques.
Picture normalization and input image noise are not
taken into account.

III. Existing Methodologies IV. Proposed Methodologies

[8] Over the last few years researches are going on regarding
vehicle identifications. Some of the studies have been done A. Binary Image Processing
just to analyze the category of vehicle for example a car, van,
 This method is used to mine license plate regions from
background. it is a combination of edge statistics and
morphology techniques. This process has achieved a 90 C. Contrast Extension
percent recognition rate from 100 images—supposing that the
number plate frame’s edges are perfect and plane. To expand the contrast of the image we have to perform the
process of histogram equalization. Contrast extension process
increases the sharpness of the image. Gray level histogram of
an image indicates the brightness of a pixel. Histogram
equalization is done to improve the quality of an image which
has a very poor contrast. The total process is divided in four
steps:
(i) summing up all the histogram values
(ii) dividing these values with the total no of pixels so as to
normalize the values.
A.1 Gray-Scale Processing (iii) enlarge these values with the highest grey level value.
(iv) chart the new grey level value.
Greyscale Images are those images which contain only a
single value that is each pixel has only a single value, they D. Median Filtering
carry only the information of intensity under them. They are
also known as black and white image or a monochrome image Median filter is used for removing the undesirable noises in
as they mostly in grey colour the intensity is divided in such a the image. In this method a matrix of 3x3 is passed in the
way that black has the lowest intensity while white has the image. According the noise levels these dimensions can be
strongest. We firstly start by converting an colour image into adjusted. This process involves sorting of all the pixel values
an greyscale image. The expression is: orderly, and then replacing the pixel being considered with the
median pixel value.
R=rgb2grey(p)
E. Character Segmentation
Where R is the gray-scaled image and p is the colour image
By using the Region props function of MATLAB the
A.2 Colour Processing characters of the resulted number plate region are separated
which gives us the defined boxes for each of the characters.
[11] Colour processing is a fundamental step in image The smallest defined box that contains a character is returned
processing as well as for plate recognition as in most of the by Region props function. This method is used to obtain the
countries certain norms are fixed for the plate colour and no. defined boxes of all characters in the number plate.
like in India the vehicles have to keep the letters in black with
a white background. But due to poor lighting conditions and
plate location the output is not efficient that is why we need
colour processing so as to have an accurate retrieval of
characters with greater efficiency.

B. Adaptive Thresholding

Before proceeding with thresholding the images must be

converted in greyscale. Thresholding is done so as to create a
binary images. Adaptive thresholding is a process in which a
threshold value is calculated and then each pixel is compared
with that constant(threshold) value and replaced with a pixel
of black color if the value is less than the constant value or a
white pixel if the value is greater than the constant value.
On the basis of local mean of pixels intensity, the adaptive
threshold is formulated: Fig1.Masking Images

O(X,Y)= 255 I(X,Y) < α+β F. Feature Extraction

O(X,Y)= 0 I(X.Y) > α-β
[12] In Feature extraction process we find, we mark, and save
Where I and O are the input and output images respectively. all the features from the number plate segmented. To
The window size parameters, m and n, are chosen based on recognize the character in number plate images we use zonal
the characters size in the region. density feature. In Zonal density function image is divided
into different areas and object’s pixel in each of the area is format of standard character to coordinate the objective
been counted. The density of each area is the total object’s character from eight bearings of up, down, left, right, upper
pixel. Total area in the image equal to total features acquired left, bring down left, upper right, bring down right.
in the
image. For 16 zonal density we divide a 32x32 image, so
that in an image there are 16 features. In order to be divided
into 16, 64, 128, 256 zones the pixel should be 32 x 32.

Fig. 2. Recognition

Fig. 1. Cropped Image

G. OCR by use of Template Matching

[6] One of the Character Recognition techniques is template

matching. It's miles the procedure of locating the region of a
sub-photograph called a template, inside an picture. Matching
of templates entails figuring out resemblances between a
given template and home windows of the same
size in an image and figuring out the window that produces Fig. 4. Template creation
the very best similarity degree. it works by comparing each
and every pixel of the photograph and template for every
feasible template displacement. This method involves the use Actual Predicted Mismatched Accuracy
and help of a database of characters or templates. For all Plate Plate Characters
feasible input characters there exists a template. for every TN 09 AX TN 09 AX 0 100%
alphanumeric characters templates are created (from A-Z 3100 3100
and zero-nine) the use of 'regular' font style. figure 3
demonstrates the templates for few of the alphanumeric
characters. TN 11 K TN 11 K 0 100%
3613 3613
[9] For acknowledgment to take place, the present information
character is contrasted with every format to discover either a TN 11 Z TN 11 2 1 85%
feasible match, or the layout with the nearest portrayal of the 3612 3612
information character. It can catch the ideal position where the
character is by moving standard layout, in this manner do the
correct match. Moving the layout coordinating technique
depends on the format of the target character, utilizing the
 VI. Conclusion

[10] The number of automobiles on the road nationwide and

in the automotive industry have grown dramatically during the
past ten years. Tracking automobiles based on their license
plate is required for better vehicular traffic management. After
the rigorous study of the available literature it can be inferred
that various mechanism have been developed for number plate
detection. Real-time object detection, image processing,
automatic license plate recognition are easily detecting the
Fig.Licensed Number Plates used for matching of templates. vehicle number plates from videos and images as result. This
is a best way of observance of traffic than a standard approach.
V. RESULTS AND ANALYSIS
VII. REFRENCES
[5] To measure our method and precision we tend to perform
our experiment on several prototypes of vehicles with entirely [1] XiaojunZhai, FaycalBensaali, “Standard Definition
different forms, and dimensions below changing conditions. ANPR System on FPGA and an Approach to Extend it to HD”
The method of segmentation did not produce desired results in 2013 IEEE GCC Conference and exhibition, November 17-
for plates at an associated degree and plates at the edge of 20, Doha, Qatar. pp.214
picture taken, this confined the accuracy of the algorithm. [2] H. ErdincKocer and K. KursatCevik, "Artificial neural
networks based vehicle license plate recognition," Procedia
Computer Science, vol. 3, pp. 1033-1037, 2011
Actual Predicted Accuracy by Accuracy [3] A Roy and D.P Ghoshal, "Number Plate Recognition
Plate Plate clustering By for use in different countries using an improved
segmentation," in 2nd National Conference on Emerging
character Trends and Applications in Computer Science(NCETACS),
TN 09 AX TN 09 AX 92% 100%
2011, pp. 1-5
3100 3100
[4] FikriyeÖztürk and FigensÖzen, "A New License Plate
Recognition System Based on Probabilistic NeuralNetworks,"
TN 11 K TN 11 K 33 69% 100% Procedia Technology, vol. 1, pp. 124-128,2012
3613 [5] Anton SatriaPrabuwono and Ariff Idris, "A Study of
Car Park Control System Using Optical Character
Recognition," in International Conference on Computer and
KA 19 PZ KA 19 P2 55% 85% Electrical Engineering, 2008, pp. 866-870
[6] Ch. Jaya Lakshmi, Dr. A. Jhansi Rani, Dr. K. Sri
Ramakrishna, and M. Kanti Kiran, "A Novel Approach for
Indian License Recognition System," International Journal of
Advanced Engineering Sciences and Technologies, vol. 6, no.
1, pp. 10- 14, 2011
[7] Jianbin Jiao, Qixiang Ye, and Qingming Huang, "A
configurable method for multi-style license platerecognition,"
Pattern Recognition, vol. 42, no. 3, pp. 358-369, 2009
 [8] Zhigang Zhang and Cong Wang, "The Research of [11] SaimaRafique, Mahboob Iqbal and Hafiz Adnan Habib,
Vehicle Plate Recognition Technical Based on BP Neural “Space Invariant Vehicle Recognition for Toll Plaza
Network," AASRI Procedia, vol. 1, pp. 74- 81, 2012 Monitoring and Auditing System”, Multitopic Conference,
[9] Ying Wen, "An Algorithm for License Plate recognition 2009. INMIC 2009,IEEE 13th International, pp. 1-6
Applied to Intelligent Transportation System", IEEE [12] SaeidRahati, ReihanehMorvejian, Ehsan M. Kazemi
Transactions of Intelligent Transportation Systems. pp. 1-16, and Farhad M. Kazem “Vehicle Recognition Using Contourlet
2011 Transform and SVM,” Proceedings of the Fifth International
[10]Chirag Patel, Dipti Shah, Atul Patel," ANPR: A Conference on Information Technology, 2008
Authorized licensed use limited to: MIT-World Peace [13] XiaojunZhai, FaycalBensaali and Reza Sotudeh,
University. Downloaded on March 24,2023 at 06:21:53 UTC “OCR-Based Neural Network for ANPR” in IEEE, 2012. Pp1
from IEEE Xplore. Restrictions apply. International [14] Y. Amit, D. Geman, and X. Fan, “A coarse-to-fine
Conference on Advances in Computing, Communication strategy formulticlass shape detection,” IEEE Transactions on
Control and Networking (ICACCCN2018) 843 Survey", Pattern Analysis and Machine Intelligence, vol. 26, pp. 1606-
International Journal, 2013 1621, 2004