Vous êtes sur la page 1sur 5

Ch.Jaya Lakshmi et al.

/ (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES


Vol No. 6, Issue No. 1, 010 - 014

A Novel Approach for Indian License Plate


Recognition System
Ch.Jaya Lakshmi, Dr.A.Jhansi Rani Dr.K.Sri Ramakrishna M.KantiKiran
V.R.Siddhartha V.R.Siddhartha V.R.Siddhartha V.R.Siddhartha
Engineering College Engineering College Engineering College Engineering College
Vijayawada Vijayawada Vijayawada Vijayawada
jayavrsec@gmail.com jhansi9rani@gmail.com srk_kalva@yahoo.com kantikiran81@googlemail.com

T
Abstract - License Plate Recognition System (LPRS) is one of works in three steps, the first step is the detection and
capturing a vehicle image, the second steps is the detection
the most important part of the Intelligent Transportation and extraction of number plate in an image. The third
System (ITS). License plate automatic recognition has been section use image segmentation technique to get individual
extensively applied in the Intellectual traffic system and character and optical character recognition (OCR) to
recognize the individual character with the help of database
Public transit security system. The main objective is to design stored for each and every alphanumeric character.
ES
an efficient automatic authorized vehicle identification system
by using the vehicle number plate. Location of the license
plate is performed using its inherent texture characteristics
In India, Vehicles used for domestic purpose contains the
license plate with white color as its background and black
characters as its foreground whereas vehicles used for
and wavelets. Character segmentation is done using
commercial purpose contains the license plate with yellow
Connected Component Analysis. Character Recognition of the color as its background and black characters as its
vehicle license plate based on template matching. This system foreground.
is implemented using MATLAB. The proposed system mainly
II. Related work
applicable to Indian License Plates.
A
There are several common algorithms to locate the license
Keywords : LPR, Wavelets, Location, Segmentation, Character plate. Searching algorithms mainly rely on color
Recognition. information and special signs. Widely used procedures that
are solely based on image processing are as Hough
transform, Top-Hat and Bottom-Hat filtering (highlights
I.INTRODUCTION
the black-white transitions) [3] Binary morphology
IJ

algorithm (for example: classical Otsu method) [5]Edge


The Automatic Number Plate Recognition (ANPR) was finding methods (Sobel, Laplacian, Roberts, Prewitt,
invented in 1976 at the Police Scientific Development Canny operators) [4] [5] [2] Procedures based on the color
Branch in the UK. License Plate Recognition system of the background and characters. Region-growing
(LPRS) is similar to ANPR. However, it gained much algorithm (RGA): By using a recursive region-growing
interest during the last decade along with the improvement algorithm, the dark regions (license plate symbols)
of digital camera and the increase in computational surrounded by light areas (background of the license plate)
capacity. It is simply the ability to automatically extract can then be classified. Each region has a unique position
and recognition of vehicle number plate’s characters from and dimensions. [1]Checking: color, size, ratio.
an image. In essence it consists of a camera or frame
grabber that has the capability to grab an image, find the Presently, there are several common algorithms for the
location of the number in the image and then extract the segmentation of license plate characters, such as direct
characters for character recognition tool to translate the segmentation, template matching, projection and cluster
pixels into numerically readable character. It can also be analysis. (1) In the direct segmentation algorithm[6], the
used to detect and prevent a wide range of criminal license plate characters are segmented directly according to
activities and for security control of a highly restricted the prior knowledge of the width of characters and spaces
areas like military zones or area around top government between characters. This algorithm is simple and rapid. But
offices. The system is computationally inexpensive the left and right side of characters region must be located
compare to the other ANPR systems. The ANPR system

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 10


Ch.Jaya Lakshmi et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 6, Issue No. 1, 010 - 014

accurately before using it. So it is reliable to the effect of IV.IMPLEMENTATION


license plate location. (2) In the template matching
algorithm [7], a template is used to scan the image of A. Image Acquisition
license plate to find the maximum difference value of the
number of white points between the region of character and The first step is the capturing of an image using the USB
the region of space between characters. This algorithm can camera connected to the PC. The images are captured in
avoid falsely segmenting characters.(3) In the projection RGB format so it can be further process for the number
algorithm with template matching[8] [9], the number of plate extraction. Pre-processing of the captured image is
white points in vertical direction is counted and recorded. performed such as RGB to gray scale conversion, noise
The character region has more white points than the region filtering, Binarization process.
of space between characters. By detecting the trough of
white points between characters, the spaces can be located B. Location of License Plate
and then the boundary of characters can be located.
This plate localization algorithm is based on combining
Character Recognition uses the basic methods as pattern textural characteristics of license plate and morphological
recognition and Neural Network based recognition operation sensitive to specific shapes in the input image
systems.(1) In Pattern Recognition method[10], the with a good threshold value by which the license plate is
character can be written differently so the pattern may located. A fine percentage of localization of License plates
varies. this may give false result.(2) Neural Network based is achieved by this algorithm. This is a better performing

T
method[11][12], this system needs training to recognise the algorithm for License Plate Images with complicated
characters. In this computational time depends on the background. License Plate consists of many vertical edges
training set and also it is very expensive. because it consists of Borders, Characters, and Digits.
Sobel mask is used to detect vertical edges in the input
image. Wavelet decomposition is performed to have better
III. STRUCTURE OF TYPICAL LPR SYSTEM analysis. The resultant image is converted into a binary
ES
The typical LPR system consists of four major parts:
Image capture, license plate location, character
segmentation and character recognition. As shown in
Figure 1.
image. Morphological operations such as erosion and
dilation are performed to find the location of the license
plate.

C. Character Segmentation

In this paper, the segmentation of license plate characters


Image Acquisition is done after license plate location and binarization. A
figure of license plate region after license plate location
and binarization is displayed as Fig. 2. Suppose that the
image data after license plate location and binarization is
stored in matrix B. The number of columns of BW is stored
A
in C. The number of rows of BW is stored in R.
Location of License Plate

Character Segmentation
IJ

Figure 2. License Plate after Location


Character Recognition
1) Precise Location of the Top and Bottom Boundary

In this step, a scanline algorithm is used to accurately


locate the top and bottom boundary of license plate
Figure 1. Vehicle license Plate character recognition characters. The scanline algorithm is based on the feature
system block diagram that the gray scale in character region changes frequently.
Thus the total number of transition in character region is
more than the total number of transition in other region.
There are atleast seven characters in license plate region

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 11


Ch.Jaya Lakshmi et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 6, Issue No. 1, 010 - 014

and every character has more than two Jumps. So we can


choose fourteen as a threshold value. To be on the safe
side, we can choose twelve as the threshold value. If the
total number of transitions in a certain line is greater than
twelve, this line may be in character region. Otherwise, it is Figure.4. License Plate after precise location
not in character region.
The specific algorithm for precise location of the top and 2) Vertical Projection and Thresholding
bottom boundary is as follows.
Project the figure in vertical direction after precise location
(a) Location of Top boundary of the top and bottom boundary. Because black points in
character region is much more than the black points in the
Scan the figure of license plate from the half of it to the top region between characters, the result of projection appears
of it line by line. If the total number of Jumps in a certain continuous peak-trough-peak.
line is less than twelve, it can be marked as the top The specific algorithm for vertical projection and
boundary of character region. thresholding is as follows.

(b) Locate the bottom boundary (a) Vertical projection

Scan the figure of license plate from the half of it to the Scan the figure from left to right column by column after

T
bottom of it line by line. If the total number of Jumps in a precise location of the top and bottom boundary and count
certain line is less than twelve, it can be marked as the the total number of black points in every column. The
bottom boundary of character region. result is stored in array projection . It is obvious that the
length of Project is C.

(b) Thresholding
Input B
ES Suppose that one represents peak and zero represents
trough. The threshold value is set to h/10. Judge every
value in array projection . If Projection[i] is greater than
h/10, Projection[i] is set to one. Otherwise, Projection[i] is
set to zero. Where h is the modified no of rows of the
Row = R/2 binary image after precise location of top and bottom
boundaries.

Count Transitions
A
from 0 to1 and 1 to
0 and store it as n Row=Row+1 Figure.5. Character segmentation

D. Character Recognition using template matching

Y In order to overcome the shortcomings of the simple


IJ

n>=12
template matching algorithm, low-resolution template
matching method is adopted, namely the using of a lower
N pixel resolution (such as 4 x 8) to represent the images and
templates to be recognized. Each matrix element
Bottom=Row corresponds to a sub-matrix in a high-resolution matrix.
The element's value is the average of the pixel gray value
in the corresponding high-resolution sub-matrix. Compared
with the high-resolution matching algorithm, correct
Figure.3. Algorithm for location of bottom boundary. identification rate of the letters and numbers is greatly
enhanced. The reason is that if the resolution rate goes
through a moderate reduction, the error generated by the
image distortion and the noise will be decreased. The
recognition errors of letters and numbers mainly occur in

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 12


Ch.Jaya Lakshmi et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 6, Issue No. 1, 010 - 014

some of the characters with the very similar main structures


but some detailed differences, such as C and 0, 3 and 5.
When we use the similarity method to calculate the
matching degree, the evaluation of these values provided
by these subtle differences is crucial. The use of high-
resolution template matching method, these deviations in
the relevant calculation have a serious interference of the
evaluation of the value function, which makes the
calculation unstable. When we use the low-resolution
template matching method, the deviation of this alignment
will have a small proportion in each element of a matrix.
Therefore, in a considerable number of cases, it does not
affect the structure description of the alpha-numeric and
thereby increases the correct identification rate.

1) Calculation of Similarity

The similarity is function which is used to measure the


degree of similarity between the patterns to be identified

T
and templates. In the template matching method, the
formula of similarity has important implications on the
recognition results. In the low-resolution template
matching method, using the similarity discriminate method
mainly include two kinds: absolute distance and similarity.
The absolute value distance regards the sum of the distance
between image and the corresponding image pixel value of
ES
template as a function to measure the similarity.

Given that image F and the template M are respectively


image of X x Y, the formula to calculate the absolute value
of the distance is as follows:

X 1 Y 1
D( F , M )  F (i, j )  M (i, j ) (1)
i 0 j 0

V. EXPERIMENTAL RESULTS Finally the displayed license number is KA 19 P 8488.


A
VI. CONCLUSION

The ultimate goal is to obtain vehicle information from


License Plate Recognition System and the recognition and
extraction of the license plate is the key of parts in the
IJ

vehicle monitoring processing. In this paper, it proposes


and designs the algorithm of the image processing and
segmentation, based on the analysis of the variety of
character recognition system. Through experiments and
improvements, the effect of the improvement is obvious
and the character recognition rate increased substantially.

ACKNOWLEDGEMENT

This research was supported by grant from Department of


TIFAC CORE in Telematics in V R Siddhartha
Engineering College, Vijayawada..

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 13


Ch.Jaya Lakshmi et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 6, Issue No. 1, 010 - 014

REFERENCES

[1] Optimization of vehicle licence plate segmentation and


symbol recognition, R.P. van Heerden and E C. Botha,
Department of Electrical, Electronic and Computer
engineering, University of Pretoria, South Africa

[2] Robust License-Plate Extraction Method under ComplexImage


Conditions, Sunghoon Kim, Daechul Kim, Younbok Ryu, and
Gyeonghwan Kim, Dept. of Electronic Engineering, Sogang
University, Seoul, Korea
[3] Automatic Car Plate Recognition Using a Partial
SegmentationAlgorithm, Fernando Martin, David Borges,
Signal Theory and Communications Department, Vigo
University, Pontevedra, Spain
[4] License plate recognition system, David Chanson and
Timothy Roberts, Department of Electrical and Electronic
Engineering, Manukau Institute of Technology, Auckland
[5] License Plate Recognition - Final Report, Pierre Ponce,
Stanley S. Wang, David L. Wang.
[6] Hongwei Ying, Jiatao Song, Xiaobo Ren” Character
Segmentation for License Plate by the Separator Symbol's

T
Frame of Reference” 2010 International Conference on
Information, Networking and Automation (ICINA)
[7] Deng Hongyao, Song Xiuli “License Plate Characters
Segmentation Using Projection and Template Matching”, 2009
International Conference on Information Technology and
Computer Science
[8] Shuang Qiaol , Yan Zhul , Xiufen Li l , Taihui Liu2 ,3,

[9]
ES
Baoxue Zhangl “Research of improving the accuracy of
license plate character segmentation” 2010 Fifth International
Conference on Frontier of Computer Science and Technology.
Dong- June Lee, Seong-Whan Lee” A New Methodology for
Gray-Scale Character Segmentation and Recognition”, 1995
IEEE.
[10] Yucheng Li, Yisong Liu Mushu Wang “Study and Realization
for License Plate Recognition System”,2009 IEEE.
[11] c. Nelson Kennedy Babu, Krishnan Nallaperumal “A License
Plate Localization using Morphology and Recognition”, 2008
IEEE.
[12] Baoming shan “License Plate Character Segmentation and
Recognition Based on RBF Neural Network”,2010 IEEE
A
IJ

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 14

Vous aimerez peut-être aussi