smart number plate identification using back propagation neural network
TRANSCRIPT
8/9/2019 Smart Number Plate Identification Using Back Propagation Neural Network
http://slidepdf.com/reader/full/smart-number-plate-identification-using-back-propagation-neural-network 1/5
International Journal of Application or Innovation in Engineering& Management (IJAIEM)Web Site: www.ijaiem.org Email: [email protected]
Volume 4, Issue 3, March 2015 ISSN 2319 - 4847
Volume 4, Issue 3, March 2015 Page 151
ABSTRACT
Number Plate Identification (NPI) process in profiles or broadcasting of moving images is mostly made up of the
supporters three organising methods: 1) Extraction of an authority to do plate field range; 2) Partitioning of plate
letters; and 3) Identification of all letters. This work is quite hard because of the variety of plate forms and the not
be equal external lighting qualities during profile getting. Therefore, most moves near work only under limited
conditions such as fixed illumination, limited vehicle go quickly, designated sends, and fixed backgrounds. Various procedures have been undergone growth for NPI in stationary profiles or broadcasting of moving images orders and
the reason of this paper is to classify and put a value on them. Offspring like managing schedule, mathematical
calculating capacity, and being seen value are equally directed, when accessible. At last, this paper proposes to
scientists a connection to common profile knowledge-base to define a general mention of end for NPI processing
tax.
Keywords: NPI, Extraction, Partitioning, Identification.
1.INTRODUCTION
Intelligent transportation systems (ITSs) are powered of 16 classes of knowledge-based structures separated into ready
brain base structure systems and ready brain vehicle systems. Data observation and letter identification process for NPI
are used as basic parts for ready brain structure networks like electronic remittance arrangement (number settlement
and station payment) and chief highway and blood-vessel related management techniques for traffic monitoring. NPIalgorithms are generally made up of the supporters three processing steps: 1) Placing of the Number plate (NP) area; 2)
Partitioning of the plate letters; and 3) recognition of each letter. The first two steps make into one profile managing
process on stationary profiles or frame orders (videos), whose put value is dependent on the correct identification
quantity and the error being seen rate.
NPI process should work quick enough to put into effect the importance of ITS. In trade terminology, a “same-time”
action for NPI stands for a quick-enough working to not felling the loss of a sole thing of interest that proceeds through
the place. Though, with the increasing change extension of the organizing capacity, the most recent growth works
within less than 50 ms for plate location.
2.PROBLEM DEFINITION
While the number plate identification processes operate outside, it is simply worried by illumination different in some
way or the composite feeling of material of vehicles. Unhappily, these troubles will drop the recognition accuracy. In
sequence to design a strong number plate identification method, it is necessary to get the details of troubles and two key
issues are made as following:
2.1 DISTURBANCES CAUSED BY THE VEHICLE HEADLIGHT OR RADIATOR The final execution of number plate identification would be greatly acted on by the number plate adaption, which is the
first step of procedure. Some spatial face coverings have been made to have more quality of number plate and register
histogram process to discover its position. However, the automobile headlights and heat processing apparatus
frequently allows identical qualities like a number plate and then the process could simply become feeble to give
positions of the real number plate.
2.2
DISTURBANCES CAUSED BY THE ILLUMINATION VARIATION AND COMPLEX TEXTURE OF
VEHICLES
In general condition like roadways or highways, the composite feeling of material of automobiles and the changing
illumination easily caused the got from qualities image to be not clear. The same hard question also has existence whena number plate is blackened. As an outcome, the removed letters will be over-segmented or under-segmented, which
will lower the recognition perfection of the removed letters.
Smart Number Plate Identification Using Back
Propagation Neural Network
Prof. Pankaj Salunkhe1, Mr. Akshay Dhawale
2
1 Head of Department (Electronics & Telecommunication Engineering), YTIET, Bhivpuri [MH], India
2 Pursuing Master of Engineering in Electronics & Telecommunication, YTIET, Bhivpuri [MH], India
8/9/2019 Smart Number Plate Identification Using Back Propagation Neural Network
http://slidepdf.com/reader/full/smart-number-plate-identification-using-back-propagation-neural-network 2/5
International Journal of Application or Innovation in Engineering& Management (IJAIEM)Web Site: www.ijaiem.org Email: [email protected]
Volume 4, Issue 3, March 2015 ISSN 2319 - 4847
Volume 4, Issue 3, March 2015 Page 152
3.LP DETECTION
With regard to the withdrawal of the plate area is disturbed, a grouping of techniques that were stated in the data comes
after, together with the explanation of the major organizing techniques. Some of these techniques can be arranged into
other than single sort and that partition between subclass is not always clear.
3.1 BINARY IMAGE PROCESSING
Procedures based upon groups of border practice and scientific structure presented very good outcomes. In these processes, the gradient size and the nearby contrast of a profile are worked out, established on the sense that the change
of light in the NP field is stranger and more constant than in another place. In the relating to replica situated directly
below the preferred procedure is based on the structural operation called “top-hat,” which is able to give position of
small ends of importantly different brightness.
3.2 GRAY-LEVEL PROCESSING
1) Universal Profile Processing: Comelli et al. [6] introduced in 1995 a structure (RITA) for the being seen of
automobile NPs. In phrase of profile operating for plate placing, RITA was collected of a NP area placing structure
and a pre-processing structure.
2) Partial Profile Study: Identical process appears where the automobile image is examined with N -row distance,
attaching the existence edges. If the gateway value is less than amount of border, the plate existence is expected.
3) Statistical Measurements: A Structure-based shade of gray procedures was also conferred in the data. Identical
to the process mentioned above, structures with a peak border size or peak border contrast are identified as probable NP field.
4) Hierarchical Representations: In, a careful way established on vector quantizer (VQ) to operate automobile
images is conferred. VQ converting into particular form can provide some indication regarding the contents of
profile area, allowing extra data that can be used to improve position execution.
Fig.1: (Top) SCW method. (Bottom) Emerging image following SCW completion
4.METHODOLOGY
4.1
BASIC BLOCK DIAGRAM
8/9/2019 Smart Number Plate Identification Using Back Propagation Neural Network
http://slidepdf.com/reader/full/smart-number-plate-identification-using-back-propagation-neural-network 3/5
International Journal of Application or Innovation in Engineering& Management (IJAIEM)Web Site: www.ijaiem.org Email: [email protected]
Volume 4, Issue 3, March 2015 ISSN 2319 - 4847
Volume 4, Issue 3, March 2015 Page 153
4.2
PLATE EXTRACTION
To increase class of automobile images for superior results in further performance on the Number plate, Image Pre-
processing actions are performed on the original car image. “A License Plate Position Method Based on Tophat-Bothat
Changing and Line Scanning”, establishes the entire action of image pre-processing and of plate region recognition.
Figure 2: Plate Extraction
4.3 CHARACTER SEGMENTATION
The Number plate acquired from Plate Extraction has letters is gray-scale. To get segmented letters, first plate image ischanged into binary image. Then 'Lines' purpose is used to separate wordings on the number plate into lines, which
uses “clip” role. “Clip” role crops black letter with white back. After taking away unwanted parts of image, resizing is
done and same action is redone on the cropped image. This procedure is moved after till all the letters are segmented.
Figure 3: Character Segmentation
4.4 FEATURE EXTRACTION
In this paper, as Neural Network is used for letter recognition, Feature Extraction is a main step for training and acting
the Neural Network. Feature Extraction is executed on each segmented letter. Two feature extraction processes are tried
for training and acting Neural Network. One is Fan-beam Transform and other way of doing is based on Character
Geometry.
4.4.1 CHARACTER RECOGNITION
This step is the major part of the structure and is called as Character Recognition step, where segmented letters areidentified. Character Recognition is also called as Optical Character Recognition (OCR).
4.4.2 BACK PROPAGATION NEURAL NETWORK
BP (Back Propagation) Neural Network is an oversaw neural network, with three layers input layer, output layer and
hidden layer. Learning rate used for training the BPNN is 0.09 and transfer function is tansig.
8/9/2019 Smart Number Plate Identification Using Back Propagation Neural Network
http://slidepdf.com/reader/full/smart-number-plate-identification-using-back-propagation-neural-network 4/5
International Journal of Application or Innovation in Engineering& Management (IJAIEM)Web Site: www.ijaiem.org Email: [email protected]
Volume 4, Issue 3, March 2015 ISSN 2319 - 4847
Volume 4, Issue 3, March 2015 Page 154
Figure 1: Original Image
Figure 2: Filtered & Equalised Image
Figure 3: Top-Hat Image
Figure 4: Threshold Image
Fig.4: Character Recognition
5.CONCLUSIONA new process has been introduced to eliminate camera shake from a single image. Established on the study-based
sparsity before of images in the framelet lands ruled over and a varied regularization on motion-blur sharpening, which
comprises both the study-based sparsity before sharpening in the framelet lands ruled over and the smoothness before
on sharpening, our new rules to make on motion deblurring has led to a strong process that can recapture a clear image
from a given motion-blurred image.
8/9/2019 Smart Number Plate Identification Using Back Propagation Neural Network
http://slidepdf.com/reader/full/smart-number-plate-identification-using-back-propagation-neural-network 5/5
International Journal of Application or Innovation in Engineering& Management (IJAIEM)Web Site: www.ijaiem.org Email: [email protected]
Volume 4, Issue 3, March 2015 ISSN 2319 - 4847
Volume 4, Issue 3, March 2015 Page 155
REFERENCES[1] T. Naito, T. Tsukada, K. Yamada, K. KO&, and S. Yamamoto, “Robust license plate recognition method for
passing vehicles under outside environment”, , IEEE Trans. on Vehicular Technology, Vol. 49, pp. 2309-2319,
2000.
[2]
Jun-Wei Hsieh, Shih-Hao Yu, and Yung-Sbeng Cben, “Morphology-based license plate detection from complex
scenes”, Proc. 16’ International Conference on Pattern Recognition, Vol. 3, pp. 176-179,Aug., 2002.[3]
Da-Shan Gao and Jie Zhou, “Car license plates detection from complex scene”, 5th International Conference on
Signal Processing Proceedings, Vol. 2, pp. 1409-1414, Aug., 2000.
[4]
S.H. Park, K.I. Kim, IC Jung, and H.J. Kun, “Locating car license plates using neural networks”, Electronics
Letters, Vol. 35, pp.1475-1477, Aug., 1999.
[5]
K. Kanayama, Y. Fujikawa, K. Fujimoto, and M. Horino, “Development of vehicle-license number recognition
system using real-time image processing and its application to travel-time measurement,’’ IEEE Veh. Techno/.,
pp. 798-804, 1991.
[6]
P. Comelli, P. Ferragina, M.N. Granieri, and F. Stabile, ‘‘Optical recognition of motor vehicle license plates”,
IEEE Trans. on Vehicular Technology, Vol. 44, pp.790-799, 1995.
[7]
M. Fahmy, “Automatic number-plate recognition: neural network approach”, Vehicle Navigation and Information
Systems Conference, pp. 99-101, Aug., 1994.
[8]
Jian-Yang, Zeng, “A research on plate localization, pre-processing and recognition of license plate recognitionsystem”, T NCTU CLENG 1999 pt.7:3
[9] Yen-Tseng Hsu, Chan-Ben Lin, Sea-Chon Mar, and Shun-Feng Su, “High Noise Vehicle Plate Recognition Using
Grey System”, Journal of Grey System, Vol. IO, No 3,pp. 193-208, 1998