Soft Comput
DOI 10.1007/s00500-014-1245-5
METHODOLOGIES AND APPLICATION
Shadow aware license plate recognition system
Shaimaa Ahmed El-said
© Springer-Verlag Berlin Heidelberg 2014
Abstract During recent years, license plate recognition
have been widely used as a core technology for security or
traffic applications such as in traffic surveillance, parking lot
access control, and information management. In this paper,
Shadow Aware License Plate Recognition (SALPR) system
is proposed to recognize Egyptian LP. This system achieves
high recognition rate through applying shadow detection and
removal, rotation adjustment and using Multilayer percep-
tron as a powerful tool to perform the recognition process.
To show the efficiency of the proposed system, experiments
have been done on numerous captured images including vari-
ous types of vehicles with different lighting and noise effects.
The experimental results yield 95.5 % recognition accuracy,
the recognition process takes 1.6 s to recognize plate infor-
mation. Most of the elapsed time used is for the license plate
extraction and rotation adjustment. The results show the fea-
sibility of the methodology followed in this paper. Perfor-
mance comparison between SALPR and other LP recogni-
tion techniques shows that for most of the cases, SALPR
performs better than other techniques under different lighting
conditions and it shows the high robustness of the proposed
algorithm.
Keywords License plate recognition · Shadow removal ·
Plate detection · Character segmentation · Skew correction ·
MLPNN · Egyptian license plates
Communicated by V. Piuri.
S. A. El-said (B )
Electronics and Communications Department,
Faculty of Engineering, Zagazig University, Zagazig, Egypt
e-mail: eng.sahmed@windowslive.com
1 Introduction
With new developments in computer and communications
technologies the need for improving Intelligent Surveillance
Systems technologies is becoming more significant. The
main role of visual traffic surveillance is capturing traffic
data, detecting accidents and safety management in general.
Learning based license plate (LP) detection methods using
different classifiers become very popular. The basic idea is to
use a classifier to group the features extracted from the vehi-
cle images into positive class (LP region) or negative class
(non-LP region). A number of computational intelligence
architectures, such as artificial neural networks, genetic pro-
gramming, and genetic algorithms, were implemented for
LPs detection. Recently, adaptive boosting (AdaBoost) and
support vector machine have been widely used for LP detec-
tion as they do not need a large number of parameters to
obtain a decent classification performance. Moving vehicles
are often extracted with their associated cast shadows after
the application of background subtraction to traffic image
sequences. This phenomenon may lead to object losses,
object shape distortion of detected vehicle. In some other
situations particularly when there are bunch of vehicles near
each other shadow of one vehicle may partially or completely
be on another vehicle and this results in misdetection of these
two or a group of separate vehicles as one big vehicle. Prob-
lems associated with occlusion would be created afterwards.
As a result the performance of the surveillance system would
be affected if the cast shadows are not detected and removed.
1.1 Shadow
Shadow detection is an important aspect of most object
detection and tracking algorithms. Shadows and shadings
in images occur when objects occlude light from a light
123