Download Automatic License Plate Recognition Draft-Computer Sciences-Project Report and more Study Guides, Projects, Research Applications of Computer Sciences in PDF only on Docsity!
Table of Contents
List of Figures .................................................................................................................... iii
iv
ABSTRACT
The name of our system is “Automatic License Plate Recognition (ALPR)”. The
objective of the software is to uniquely identify vehicles on the basis of license plate
attached to the vehicle using the front or back images of vehicles; it assumes no
additional mean of vehicle identity. The image will be automatically processed by the
Automatic License Plate Recognition System. The system will extract the number plate
from that image and recognize the number on it. Once the location of license plate is
detected, its digits and alphabets are recognized then those can be saved or checked
against a database.
In the current modern era, the use of automated and intelligent systems is becoming more
and more common. The intelligent transport system technology has gotten so much
attention that many systems are being developed and applied all over the world. License
Plate Recognition system has many applications in traffic monitoring system, including
the traffic volume, ticketing vehicle without human interruption, security and so on.
Chapter 1
Introduction
This system is able to extract various kinds of license plates but only with English characters on the license plate. A license plate (with regards to motor vehicles) is a label, usually attached to objects to specify a certain government license code. This code is unique, because it must identify the exact object (or vehicle in other words) that carries it. Car license plates are meant to be recognized by people (human beings), especially government officials, such as policemen and transport personnel. If a car is involved in robbery or stealing, for instance, then seeing its plate helps tracking down the criminals. If a stolen car is found, its license plate helps to find the owner. The system is developed only for offline recognition i.e. it will not be able to carry out license plate recognition online using a movie camera etc. Our system will work on still images captured using a digital camera, the images are taken in a natural environment and there is a lot of background information (or information which is useless to our system) is involved. The task of the application is to extract and then detect the number plate area from the specified image with a lot of extra information in it. The most crucial and the difficult part of a LPR system is the detection and extraction of the vehicle license plate, which directly affects systems overall accuracy. The presence of noise in the image, uneven illumination, dim light and foggy conditions make the task even more difficult.
LPR is also called in different references as:
Automatic Vehicle Identification (AVI) Car Plate Recognition (CPR) Automatic Number Plate Recognition (ANPR) Car Plate Reader (CPR) Optical Character Recognition (OCR) for Cars
cases the LPR unit is added as retrofit in addition to existing solutions, such as a magnetic card reader or ticket dispenser/reader, in order to add more functionality to the existing facility.
1.1 Working of a Typical LPR System
The working of an existing LPR system available in market is discussed here. The LPR application runs as a background Windows application in the PC, and interfaces to a set of camera/illumination units. The application controls the sensors and controls via an I/O card that is connected through a terminal block to the inputs and outputs. The application displays the results and can also send them via serial communication. It writes the information to local database or to optional remote databases (via the network).
The following example shows how a typical access-control system works.
Camera
Signal
Magnetic Loop detector
Figure 1 Car approaching entrance area
The vehicle approaches the secured area, and starts the cycle by stepping over a magnetic loop detector (which is the most popular vehicle sensor). The loop detector senses the car and its presence is signaled to the LPR unit.
Figure 2 Capturing the front view of vehicle
The LPR unit activates the illumination (invisible Infra-red in most cases) and takes images of the front or rear plates from the LPR camera. The images of the vehicle include the plate and the pixel information is read by the LPR unit's image processing hardware (the frame grabber). The LPR unit analyzes the image, enhances it, detects the plate position, and extracts the characters on the plate.
Figure 3 Vehicle allowed entering
1.2.2 Access Control
A gate automatically opens for authorized members in a secured area, thus replacing or assisting the security guard. The events are logged on a database and could be used to search the history of events.
In this example, the gate has just been automatically raised for the authorized vehicle, after being recognized by the system. A large outdoor display greets the driver. The event (result, time and image) is logged in the database. [10]
Figure 5 Access Control
1.2.3 Tolling
The car number is used to calculate the travel fee in a toll-road, or used to double- check the ticket.
In this installation, the plate is read when the vehicle enters the toll lane and presents a pass card. The information of the vehicle is retrieved from the database and compared against the pass information. In case of fraud the operator is notified. [10]
Figure 6 Tolling
1.2.4 Border Control
The car number is registered in the entry or exits to the Country, and used to monitor the border crossings. It can short the border crossing turnaround time and cut short the typical long lines.
This installation covers the borders of the entire Country. Each vehicle is registered into a central database and linked to additional information such as the passport data. This is used to track all border crossings. [10]
Figure 7 Border Control
1.2.7 Traffic Control
The vehicles can be directed to different lanes according to their entry permits (such as in University complex projects). The system effectively reduces traffic congestions and the number of attendants.
In this installation the LPR based system classifies the cars on a congested entrance to 3 types (authorized, known visitors, and unknown cars for inquiry) and guides them to the appropriate lane. This system reduced the long waiting lines and simplified the security officers work load. [10]
Figure 9 Traffic Control
1.2.8 Marketing Tool
The car plates may be used to compile a list of frequent visitors for marketing purposes, or to build a traffic profile (such as the frequency of entry verses the hour or day). [10]
All these applications of the License Plate Recognition project make it a profitable market product.
1.3 User Characteristics
The user of the system is not required to have any technical expertise on the image processing field. The system is user friendly and any one can use it easily but some basic knowledge of computers is also required for maximum performance.
The system is meant to check the number plate of the vehicle, so the user has to interact with the simple interface of the software just by pushing buttons, no technical expertise and long lasting experience is required to operate the Software. The system will be used for scientific, engineering, image processing and vehicle registration commercially. So, the target users of the system are educated people, but as already told non-technical users can use this software with ease of use and without any technical background.
1.4 Technical Constraint
Tool will require the following environment conditions and software to provide its functionality. Although each state has different laws concerning vehicle identification, most vehicle owners are required to properly attach and display front and rear license plates. Although video is common as an LPR acquisition method, still pictures were preferred in order to simplify acquisition and processing operations. Color pictures were taken with an Orite 6.6 Mega pixel digital camera in varying lighting conditions in Rawalpindi. All pictures were taken at a 640 x 480 pixel resolution. An effort was made to include as many states and varieties as possible. Some of the pictures were also downloaded from internet and a dataset of 100 vehicle images (with Pakistani and British license plates) was obtained.
The minimum system configuration required for our system is
800 Mhz Intel Original Processor. 256 MB RAM. Windows XP 2002. Matlab version 7.1.
Chapter 2
Literature Survey
2.1 Introduction
This chapter covers the discussion of various techniques used for license plate recognition along with their advantages and drawbacks. Several approaches may be taken towards solving this problem, each having different computational expense and success rates. It is the purpose of this chapter to explore a set of representative techniques and evaluate their performance within the LPR framework.
2.2 Literature Survey
Most of the work published on automatic license plate recognition originated in countries where license plate design follows strict standards, and there is little, if any, variety amongst samples. Therefore, many of the approaches taken towards detecting and extracting the license plate region are not applicable to our problem space. We will discuss these one by one and see which one best suite for our problem. To detect and identify the region of car license plate in the past many techniques have been used, such as
Template Matching Color Edge Detection Neural Networks Edge Detection and Hough transform Edge Statistics and Morphology.
2.2.1 Template Matching
In template matching the plate is assumed to be of fixed size such as 180x40 pixels, the characters on the plate should follow a standard format and there should be no characters other than on license plate. Template matching is used to find the corner points of the plate. If four possible corner points are found, the content of the quadrangle is checked on its spatial frequencies. Certain spatial frequencies are expected due to the characters in a plate. Only in case this frequency content confirms its presence, the four corner points are accepted as being the corner points of a license plate. In this way, a powerful license plate segmenter was obtained, which is able to indicate the exact positions of the corner points with a maximum error of only a few pixels. We can also say that it tries to match an existing template of license plate with that on the image. Template matchers are basically comparing candidates with templates, the candidate that matches best to the template wins. Characters on the plate are segmented using horizontal segmentation. The horizontal segmentation of the characters of the license plate is based on finding the spaces between them. These spaces are found by examining the maxima of the column sums of the grey values of the license plate. Then those segmented characters are also recognized using template matching. Overall recognition rate achieved using this approach is 87%.
2.2.2 Color Edge Detection
In color edge detection black or red characters with white background, or white characters with red or green back ground or green characters with white background are considered only. The RGB space of the input color image is transformed into the HSI space. License plate candidates are then determined to be those interesting areas whose sizes are large enough. The color edge detector focuses on only three kinds of edges (i.e., black-white, red-white and green-white edges). Consider a black-white edge, and suppose that the input RGB color image has been normalized into an image. Ideally, the values of a white pixel and a black pixel should be (1, 1, 1) and (0, 0, 0),
2.2.4 Edge Detection and Hough Transform
It is assumed in it that the color of license plate is other than color of car, license plates are of fixed size and follow a format and number of characters on plate is fixed. The image is first converted to grayscale and histogram equalization is applied to enhance the image. After Sobel filter is used to extract the edging image. Hough transform is applied to the binary image to extract lines from object-images. Then we look for two parallel lines, whose the contained region is considered plate candidates. The contour algorithm is also applied to verify license plate, also horizontal cross count is used for verification means the number of objects on the license plate. Overall recognition rate achieved using this approach is 92.5%.
2.2.5 Edge Statistics and Morphology
It assumes that color of license plate is other than color of car and license plates are of fixed size. Vertical edge detector is used for edge detection and in suppressing horizontal noise. Before vertical edge detection, a linear filter is used to smooth the image and apply the luminance normalization to reduce the influence of light. The connected component analysis algorithm is applied to the processed images. So we get the bounding rectangle of the object and the number of the object pixels in these rectangles. Some features of region, such as the aspect ratio (R), the area (A) and the density (D) of region are applied. Let Re denote the region of rectangles with width W and height H, then R =W / H (2.1) A=W x H (2.2)
Where N denote of the number of the object pixels in the rectangles, then D=N / (W x H) (2.3)
Then objects within certain threshold for R are accepted, and also with density in specified threshold. Also priority is given to the rectangles in the bottom of the images, and to the rectangles, which correspond to the dimension and the ratio of the standard license plate. Success rate achieved using this approach is 99%.
2.3 Conclusion
In Template Matching the size of license plate is assumed to be fixed and a template license plate is used to locate the license plate in the image but this technique fails completely if the size of the license plate is changed. This case occurs frequently if the images are not taken from fixed distance.
In Color Edge detection the color of the character and the background of license plate is assumed to be fixed, which limits the number of license plates. This technique fails due to non-uniform illumination and if the color of the background or the characters is changed.
In Neural Networks instead of locating the plate characters in the image are extracted and than recognize and for this it assumes a specific size for characters on the license plate but it fails very badly if there is some other text of almost the same font size or the font size of license plate is changed due to variance of distance in between the image capturing device and the vehicle.
In edge detection and Hough transform the plate is extracted using license plate edges and verified using Hough transform but the problem lies in generalizing the results of transform for all the test cases and its very large processing time.
The last technique that we have used is Edge Statistics and Morphology which uses the characteristics and shape of edges to locate the license plate. Usually the license plate is of square or rectangular shape. This technique is very useful for non-standard license plate and also it is invariant to the license plate size. For license plate verification it uses the smoothness constraint as license plate is a textured area as compared to other candidates such as back lights, windscreen etc.
