Mayank Finalproject

1
PROJECT REPORT ON
LICENSE PLATE RECOGNITION SOFTWARE
BY
ABHINANDAN SINGH - 08070121202 MAYANK KISHORE PATHIKRIT GUHA - 08070121224 - 08070121232
Under the Guidance of
Asst. Prof. MRUDANG SHUKLA

In partial fulfilment of
B.Tech ( E& TC )
DEGREE OF SYMBIOSIS INTERNATIONAL UNIVERSITY, YEAR 2011 - 2012 DEPARTMENT OF ELECTRONICS & TELECOMMUNICATION ENGINEERING SYMBIOSIS INSTITUTE OF TECHNOLOGY PUNE 412 115.
1
CERTIFICATE
This is to certify that the project entitled LICENSE PLATE RECOGNITION SOFTWARE submitted by _ Abhinandan Singh _ , Roll No. _ 08070121202 _, is record of
bonafide work carried out by her/him , under my guidance, in partial fulfillment of the requirement for the award of Degree of Bachelor of Engineering ( Electronics & Telecommunication ) of Symbiosis International University.
Date :
Place :
Prof. Mrudang Shukla

Guide , Department of E & TC Engineering Symbiosis Institute of Technology Pune 412 115.
Prof. Dr. Narayan Pisharoty

Head, Department of E & TC Engineering Symbiosis Institute of Technology Pune 412 115.
CERTIFICATE
This is to certify that the project entitled LICENSE PLATE RECOGNITION SOFTWARE submitted by _ Mayank Kishore _ , Roll No. _ 08070121224 _, is record of
Date :
Place :


CERTIFICATE
This is to certify that the project entitled LICENSE PLATE RECOGNITION SOFTWARE submitted by _ Pathikrit Guha _ , Roll No. _ 08070121232 _, is record of
Date :
Place :


The Project entitled
LICENSE PLATE RECOGNITION SOFTWARE
By
Abhinandan Singh - 08070121202 Mayank Kishore - 08070121224 Pathikrit Guha - 08070121232
Is approved for the degree of BACHELOR OF ENGINEERING Electronics & Telecommunication Symbiosis International University, Pune.
Examiners:
1.
______________________________________
2.
_______________________________________
Date :
Place:
5
ACKNOWLEDGEMENT
The real spirit of achieving a goal is through the way of excellence and austere discipline. We would have never succeeded in completing our task without the cooperation, inspiration ad help provided to us by various personalities. With deep sense of gratitude we express our sincere thanks to our esteemed and worthy supervisor Mr. Mrudang Shukla, Assistant Professor, Department of Electronics & Telecommunication, Symbiosis Institute of technology, Pune for his valuable guidance in carrying out this work under his effective supervision, encouragement, enlightenment and cooperation. We shall be failing in our duties if we do not express our deep sense of gratitude towards Dr. Narayan Pisharoty, Head of the Department of Electronics & telecommunication, Symbiosis Institute of Technology, Pune who has been a constant source pf inspiration for us throughout this work. We are also thankful to all the staff members of the Department for their full cooperation and help. Above all we render our gratitude to the ALMIGHTY who bestowed selfconfidence, ability and strength in us to complete this work.
Table of Contents Contents List of Figures Abstract Chapter 1. Literature Survey Chapter 2. Problem Definition 2.1 Problem Statement 2.2 Features of project Chapter 3. Introduction 3.1 Project Background 3.2 How the Vehicle License Plate System works 3.3 Objectives 3.4 Constraints 3.5 Various Types of Vehicle License Plate 3.6 Commercial Products Chapter 4 Platform / Technology 4.1 Introduction to matlab 4.2 Image Processing 4.3 Comparison Between MATLAB and Artificial Intelligence Chapter 5. Components Chapter 6. Imaging Hardware Chapter 7. Development History Chapter 8. Algorithm 8.1 Algorithm for ANPR system 8.2 Input Image from file. 8.3 Convert image into binary. 8.4 Segmentation 8.5 Number identification 8.6 Save to file in given format Chapter 9. Image Acquisition 9.1Image Cropping 9.2 Conversion of RGB Images to Binary Images
7
Page no. 10 11 12 14 14 14 15 15 15 17 18 20 22 24 24 25 26 27 28 29 34 34 34 34 34 34 35 37 37 38
9.3 Colour Image Processing Chapter 10. Image Enhancement 10.1 Imcomplement 10.2 Medfilt2 Chapter 11 Thresholding 11.1 Threshold Selection 11.2 Global Thresholding 11.3 Adaptive thresholding Chapter 12 Image Noise 12.1 Filtering 12.2 Binary Area Open(Bwareaopen) 12.3 Filtering of Digits 12.4 Identify and remove away the small connected objects Chapter 13 Image Segmentation 13.1 Feature Extraction from Digital Image 13.2 Database Chapter 14 Character Recognition 14.1 Normalization 14.2 Template Matching Chapter 15 Analysis of Result 15.1 Extracting of Individual Digits 15.2 Vehicle Quantization and Equalization 15.3 Checking and verification of the Static Chapter 16 List of Functions Used Chapter 17 Code 17.1 Algorithm to extract digits 17.2 Algorithm to normalisation of characters 17.3 Algorithm to read characters Chapter 18 Result Analysis 18.1 Experimental results 18.2 Discussion Chapter 19 Difficulties and Recommendations
8
40 42 42 42 43 43 44 44 45 45 45 45 46 48 48 49 50 50 50 52 52 53 53 55 57 57 58 58 61 61 63 65
Chapter 20 Critical Review and Reflection Chapter 21 Conclusion and Future scope Annexure A: Project Planner and Progress Report References
67 68 71 72
10
List Of Figures
3.1 Automatic Number Plate Recognition System............................................16 3.2 A car approaching a License Plate Recognition System.............................17 3.3 Indian car license plate.................................................................................18 3.4 System Block Diagram.................................................................................19 3.5 Commercial and Private vehicle license plate..............................................20 4.1 Comparison of MATLAb with other softwares...........................................24 4.2 Block diagram of Image processing system................................................25 8.1 Overview of the vehicle license plate system..............................................36 9.1 Original image of the vehicle......................................................................37 9.2 Cropped image of the vehicle.....................................................................37 9.3 Characteristic function of binary image......................................................38 9.4 Histogram of light object with dark background.........................................39 9.5 Histogram of object and background which is close...................................39 9.6 RGB schematic block diagram....................................................................40 9.7 Captured image of the vehicle.....................................................................41 9.8 Binary image of the vehicle.........................................................................41 12.1 Binary image with noise.............................................................................47 12.2 Binary image after filtering........................................................................48 13.1 Segmented image.......................................................................................50 15.1 Captured image of vehicle1.......................................................................53 15.2 Captured image of vehicle2.......................................................................53 15.3Morphologically open binary image...........................................................54 15.4 Morphologically open binary image...........................................................55
10
11
ABSTRACT
Automatic Vehicle Identification (AVI) has many applications in traffic systems (highway electronic toll collection, red light violation enforcement, border and customs checkpoints, etc.). License Plate Recognition is an effective form of AVI systems. In this study, a smart and simple algorithm is presented for vehicles license plate recognition system. The proposed algorithm consists of three major parts: Extraction of plate region, segmentation of characters and recognition of plate characters. For extracting the plate region, edge detection algorithms and smearing algorithms are used. In segmentation part, filtering and some algorithms are used. And finally statistical based template matching is used for recognition of plate characters. The performance of the proposed algorithm has been tested on real images. Based on the experimental results, we noted that our algorithm shows superior performance in car license plate recognition. This system is based on the image processing system. This system helps in the functions like detection of the number plates of the vehicles, processing them and using processed data for further processes like storing, allowing vehicle to pass or to reject vehicle. This system also helps to conduct the graphic images of the vehicles which can be further stored in the database in text format reducing size of data to be stored. The general algorithm involves the following steps: Image Processing : The image captured is pre-processed and reduction in the contrast. Plate localization and extraction: To obtain the vehicle plate sub image. Character Segmentation/Recognition: Resample and threshold in order to isolate the license plate and vehicle license plate character. We used neural network for recognition of vehicle license late character. The neural network will be trained off-line with the characters and numbers. Evaluating the performance of the algorithm and compare the performance with other reported work. Implementing a file management system or database for storing the images of vehicle license plate, numbers and characters Keywords Character recognizer license plate recognition Segmentation Cropping
11
template matching Median Filter Connected Components Masking.
12
1 LITERATURE SURVEY
It is almost known that the world is nowadays being more dependent in the computer than the human manpower. Today in the developed countries more than 90% of the job is done by automated system and this is due to the accuracy, reliability and versatile nature of the computer. In the developed western countries many big factories, nuclear plants, communication centers all of them are under the control of the computer because thats not the place where human hand can reach up. To add the further mile for this step we have presented a License plate recognition software .The project is based on image processing technology using popular software MATLAB. Vehicle License Plate Recognition is an image processing system whereby it is used to recognize the vehicles by identifying the license plate. It is basically use for traffic and security purposes. The problem of automatic VNP recognition is being studied since the 90s. The early approaches were based on characteristics of boundary lines. The input image being first processed to enrich and enhance boundary line-information by using such algorithms as the gradient filter, and resulting in an image formed of edges. The image thus processed was converted to its binary counterpart and then processed by certain algorithms, such as Hough transform, to detect lines. Eventually, couples of 2-parallel lines were considered as a plate-designate. Another approach was based on the morphology of objects in an image. This approach focuses on some salient properties of vehicle plate images such as their brightness, contrast, symmetry, angles, etc. Due to these features, this method could be used to detect the similar properties in a certain image and locate the position of number plate regions. The third approach was based on statistical properties of text. In this approach, text regions were discovered using statistical properties of text like the variance of gray level, number of edges, edge densities in the region, etc. This approach was commonly used in finding text in images, and could well be used for discovering and designating candidate number plate areas as they include alphabets and numerals. All of the systems discussed above have some kind of limitations for example they are plate size dependent, color dependent, work only in certain conditions or environment like indoor images etc. License plate identification/recognition (LPI/R) is one form of ITS technology that not only recognizes and counts vehicles, but distinguishes each as unique. For some applications, such as electronic toll collection and red-light violation
12
13
enforcement, LPI/R records a license plates alphanumeric so the vehicle owner can be assessed the appropriate toll or fine. In others, like commercial vehicle operations or secure-access control, a vehicle's license plate is checked against a database of acceptable ones to determine whether a truck can bypass a weigh station or a car can enter a gated community or parking lot. License plate recognition (LPR) is a new tool for automatic vehicle and traffic monitoring by using digital image processing. For implementing LPR system we have used digital image processing technique. The LPR system can be used to traffic control management for recognize vehicles that commit traffic violation, such as entering restricted area without permission ; occupying lanes reserved for public transport, crossing red light, breaking speed limits ; etc.
13
14
2 Problem Definition 2.1 Problem statement:

Automatic car license plate recognition (CLPR) system is implemented to help the human to automatically detect plate number without human supervision. Previously, human is needed to observe and list the user car plate number manually. So this project is developing to replace human to monitor the car and automatically capture the image. Besides that, the system can automatically display the status of the car which it will compare between the car plate numbers recognized with the database.
2.2 Features
Automatic number plate recognition (ANPR) is a mass surveillance method that uses optical character recognition on images to read vehicle registration plates. They can use existing closed-circuit television or road-rule enforcement cameras, or ones specifically designed for the task. They are used by various police forces and as a method of electronic toll collection on pay-per-use roads and cataloging the movements of traffic or individuals. ANPR can be used to store the images captured by the cameras as well as the text from the license plate, with some configurable to store a photograph of the driver. Systems commonly use infrared lighting to allow the camera to take the picture at any time of the day.[ANPR technology tends to be region-specific, owing to plate variation from place to place. Concerns about these systems have centered on privacy fears of government tracking citizens' movements, misidentification, high error rates, and increased government spending.
14
15
3 INTRODUCTION
3.1 Project Background
Massive integration of information technologies into all aspects of modern life caused demand for processing vehicles as conceptual resources in information systems because a standalone information system without any data has no sense, there was also a need to transform information about vehicles between the reality and information systems. This can be achieved by a human agent, or by special intelligent equipment which is be able to recognize vehicles by their number plates in a real environment and reflect it into conceptual resources. Because of this, various recognition techniques have been developed and number plate recognition systems are today used in various traffic and security applications, such as parking, access and border control, or tracking of stolen cars. Vehicle License Plate Recognition is an image processing system whereby it is used to recognize the vehicles by identifying the license plate. In entrance gate, number plates are used to identify the vehicles. When a vehicle enters an input gate, number plate is automatically recognized and stored in database and black-listed number is not given permission. When a vehicle later exits the place through the gate, number plate is recognized again and paired with the first-one stored in the database and it is taken a count. Automatic number plate recognition systems can be used in access control. For example, this technology is used in many companies to grant access only to vehicles of authorized personnel. In some countries, ANPR systems installed on country borders automatically detect and monitor border crossings. Each vehicle can be registered in a central database and compared to a black list of stolen vehicles. In traffic control, vehicles can be directed to different lanes for a better congestion control in busy urban communications during the rush hours. Automatic Line Tracking Robot (ALR) is used in this project as a vehicle which contains circuitry for moving in a guided track. It will have mechanism to detect the opened and closed door. It also will have capacity to park in the given parking area.
.
3.2 How the Vehicle License Plate System works

Firstly, the vehicle will stop at the car gantry. The cycle will start when the vehicle steps over the detector. It will activate a signal to the Vehicle License Plate System of the presence of the vehicle.
15
16
Secondly, illumination (infra-red) will be activated and images of the front picture of the vehicle will be taken. The system will read the information pixels of the vehicle and run the recognition process. Thirdly, the system will apply certain algorithm to analyses the vehicle image. Besides analyzing, the images will be enhance, locating the vehicle plate position and extract the characters from the vehicle plate. Next, the characters will be recognized by using Neural Network. Lastly, the system will try to match the recognized vehicle plate number with the car plate database. If Access Granted, the gantry will open and allowed the vehicle to pass through. Diagram is illustrate in Fig 1.
Fig 3.1 Besides, the Vehicle License Plate Recognition also provide an advantage by keeping the image of the vehicle in which it will be useful for crime fighting. Camera can also focus on the face of the driver and save it for security reason. There are difficulties for Vehicle License Plate Recognition in which it will affect the efficiency and accuracy of the system. It is essential and important to determine the facts which will able to influence the operations and recognition proficiency. Next, we also need to look into other facts of variables that are not constant. Below are the non-constant variables which will affect the accuracy of recognition: Speed of the vehicle
16
17
Weather condition Type of Vehicle Distance between vehicle license plate and the camera Type of plate ( Rectangular, Bent type) Vehicle license plate orientation Type of vehicle fonts character.
Fig 3.2 A car approaching a License Plate Recognition System
3.3 Objectives
The overall objective of the project is to develop a system to recognize vehicle license plate from a car at a gate entrance of a parking lot. The software could lead to a cheaper and faster way of enhancing and determined the performance of the recognition system. The system will be based on a Personal Computer such that it will generate report on the vehicle license plate it has captured. Once the vehicle license plate is captured, the characters will be recognized and displayed on the Graphical User Interface. Besides, the system can also serve as a security purpose whereby it can spot on any wanted or stolen vehicles. In the past, there has been similar project implemented but had poor accuracy. Thus, we would need to improve or rewrite the algorithm to improve the accuracy. I
17
18
will address a set of constraints and focus on the design of the algorithm to extract the vehicle license plate in order to improve the accuracy. There is definitely a lot more room for further improvement on this project. However, due to the limited time frame given, it is not advisable for me to cover all aspects in this project. Thus, in discussion with my tutor, we managed to come up with the progress guideline as to what need to be included in this project.
3.4 Constraints
Due to limited time that we possess and dealing with image vision software, it is not advisable to include all of the possible cases. Thus, we have to set a list of constraints to make the project more systematic and manageable. The constraint is listed as below: Image taken only when vehicle is stationary. Captured image of vehicle at fixed distance. Captured image of vehicle at fixed angle. There will be no motion capture image. The vehicle license plate position should be captured centred. The image should be taken with the height of 50cm to 70 cm above the ground level. Take only the front view image of the car. Try on zoom in image of the car and image consists of headlamp. Captured images on location where light is proportional. Deal with only Indian Car License Plate with Times new roman font.
Fig 3.3
18
19
System block Diagram
Fig 3.4
19
20
3.5 Various Types of Vehicle License Plate What is Car license Plate
Car license Plate is a rectangular metal plate which consists of a combination sequence of alphabet characters and numbers issued by the government for identification purpose of registered vehicle.
Format
Vehicle owner who register the car as private vehicle have a white background and black text. Commercial vehicles such as taxis and trucks have a yellow background and black text (e.g., DL 2C 0001). Vehicles belonging to foreign consulates have white lettering on a light blue background (e.g. 22 UN 14). The President of India and state governors travel in official cars without licence plates. Instead they have the Emblem of India in gold embossed on a red plate.
Current Format
The current format of the registration index consists of 3 parts

The first two letters indicate the state to which the vehicle is registered. The next two digit numbers are the sequential number of a district. Due to heavy volume of vehicle registration, the numbers were given to the RTO offices of registration as well. The third part is a 4 digit number unique to each plate. A letter(s) is prefixed when the 4 digit number runs out and then two letters and so on.
This scheme of numbering has some advantages:

the State or District of registration of a particular vehicle In the case of a police investigation of an accident or vehicle-related crime, witnesses usually remember the initial area code letters it is then quite simple to narrow down suspect vehicles to a much smaller number by checking the database without having to know the full number.
Fig 3.5
20
21
Military vehicles
Military vehicles have a unique numbering system unlike any other licence numbers. The numbers are registered by the Ministry of Defence in New Delhi. The first (or the third) character is always an arrow pointing upwards, this prevents the number being read wrongly in case the plate (and/or the vehicle bearing it) is upside down. The next two digits (or the two preceding the arrow) signify the year in which the Military procured the vehicle. The next is the base code, followed by the serial number. The letter ending after the serial number indicates the class of the vehicle.
Diplomatic plates
Vehicle belonging to foreign missions bear the plates UN, CD or CC, which stand for United Nations, Diplomatic Corps or Consular Corps respectively. A diplomatic plate numbered 13 CC xxxx would refer to country 13, probably a country close to the letter A or B. For example, a vehicle bearing the number 77 CD xxxx in India refers to a vehicle owned by either the United States mission in India or by a person working with the mission.
Other numbering
Other numbering includes the special numbers allotted to public transportation such as buses, taxis and auto-rickshaws. The numbers are registered by the organization which run the services and are usually printed on the side of the vehicle. This mechanism is used for unique identification.
Temporary numbers
As soon as a vehicle is purchased, the dealer of the vehicle issues a temporary licence sticker known colloquially as a TR (To Register) number. This is valid for one month, during which the owner must register the vehicle in the controlling RTO of the area in which the owner is residing to get a standard licence plate. To register a vehicle, it has to be presented to the RTO's office, where a Motor Vehicle Inspector will verify the applicant's address and other details, confirm that the engine and chassis numbers are identical to what is written in the application and issues a permanent registration certificate which is usually valid for 20 years. The permanent registration certificate is one of the four important documents a vehicle plying on the road should always have; the others being a valid insurance certificate, a pollution under control (PUC) certificate and the driver's licence. For public utility vehicles like buses, trucks, taxis and pick-up vans, there are a number of additional documents like a roadworthiness certificate and a transportation permit.
21
22
3.6 Commercial Products

The various products in the market today are described briefly below.
3.6.1 IMPS (Integrated Multi-Pass System)

An IMP [26] is a Singaporean commercially developed license plate recognition system. It is a high performing robust system that gives consistent results under all weather conditions. Using advanced image processing and artificial intelligent techniques such as AI best first breadth-wise search algorithm, combined template and neural network recognizers, fuzzy logic and an arsenal of image processing tools, it automatically locates vehicle license plates and reads the numbers accurately each time every time.
3.6.2 Perceptics
Perceptics [27] is the world leader in license plate reader technology. Current LPR system read Latin (A-Z) and Korean (Hangul) letter and Arabic number (09); however, the LPR can be programmed to read any language or symbol in any alphanumeric combination or context on both retro and non-retro reflective plates. With milliseconds the LPR system locates, captures and identifies a vehicle's license plate data and makes a read decision. The system's reliability and flexibility allow it to accommodate some of the most stringent needs in some of the worst conditions. Features of this LPR technology includes. Automatic and within milliseconds. Reads accurately in most weather conditions. Reads accurately at highway speeds. Works 24 hours a day, 7 days a week. 3.6.3 Vehicle Identification System for Parking Areas (VISPA) PPI's Vehicle Identification System for Parking Areas (VISPA) [28], uses video imaging for better recognition, identification and improved security. VISPA provides for state-of-the-art video technology, easy installation and has accessories and features for most parking security surveillance needs.
22
23
Features are Open architecture to most common video-systems. Compatible with standard hardware and software. Can be customized according to specific user needs. VISPA is available in two forms Basic Version: - An image of the car and/or the driver (depending on the location of your camera) will be taken as soon as the car approaches the triggering device. The image will be linked to the ticket. The basic system version connects to 4 cameras and can be upgraded to 8 cameras. Enhanced Version:- License Plate Identification, The VISPA controller with an integrated frame grabber card for 4, 8, or 16 cameras automatically identifies the license plate from the video image and stores it in a database. The license plate can then be encoded on the ticket.
3.6.4 Hi-Tech Solution

Hi-Tech Solutions [29] is a system and software company that develops cutting edge optical character recognition (OCR) solutions by implementing the company's unique image processing software and hardware in a wide range of security and transportation applications. There technology is based on computer vision, the system read the camera images and extract the identification data from the images. The recognition result is then logged together with the images. This is the main advantage of vision based recognition, the records include both the image plus the extracted result. There product includes, SeeCar License Plate Recognition:- Detects and reads Vehicle license plates for parking, access control, traffic surveillance, law enforcement and security applications. Available as a complete system which is based on a background Windows application, Windows DLL or Linux library, as a stand-alone turn-key version, or in form of different special-task systems. SeeContainer Identification System:- Tracks and reads Shipping container identification marking, and transmits the ID string to the port or gate computer, or to a client process. Available as complete systems, such as SeeGate - a recognition system for the Tracks and Containers, or SeeCrane - crane mounted Container recognition system.
23
24
4 Platform/Technology
4.1 INTRODUCTION TO MATLAB MATLAB is chosen as the main development for the license plate recognition application. The task is to construct the algorithm and recognizes can be done using MATLAB. Besides, MATLAB is also very efficient as it have built-in-function tools for neural network and image processing. The advantages of MATLAB can be classified as follows: Platform independence, predefined function and deviceindependent plotting, Graphical User interface and MATLAB compiler. Comparison of MATLAB with other program is shown in the below diagram.
Fig 4.1
MATLAB read images using the function imread. The table below show images/graphs format that is supported by imread with MATLAB.
24
25
Format Name TIFF JPEG GIF BMP PNG XWD
Description Tagged image file format Joint photographs Expert Group. Graphic Interchange Format Windows bitmap Portable network Graphics X Window Dump
Extension .tif; .tiff .Jpg ; .Jpeg .gif .bmp .png .xwd
4.2 Image Prcessing

Image Processing generally involves extraction of useful information from an image. This useful information may be the dimensions of an engineering component, size of diagnosed tumour, or even a 3D view of an unborn baby. The main areas of application of Image Processing are Bio-Medical, Engineering, Quality Control, Face Detection, Traffic Control etc.
Block Diagram of Image Processing Steps
Fig 4.2
25
26
4.3 Comparison Between MATLAB and Artificial Intelligence
Advantage of Artificial Intelligence:

1. Good for solving complex problems where humans or programmer have expert knowledge in it. 2. Non-programmers can easily understand the rules of artificial intelligence. 3. We can changed and add individual rules.
Disadvantage of Artificial Intelligence:

1. There are difficulty in getting expert knowledge and putting it in used in the system. 2. Not good in common sense reasoning. 3. Expensive. 4. Slow response. 5. Unpredictable.
Advantage of MATLAB
1. Data can be represented in terms of matrices or vectors. 2. MATLAB coding is shorter and simpler. 3. Algorithm used in MATLAB can be converted to use in hardware. 4. The system will perform faster when using Matrix Approach. 5. MATLAB possess power graphic visualization tools.
.
26
27
5 COMPONENTS
The software aspect of the system runs on standard home computer hardware and can be linked to other applications or databases. It first uses a series of image manipulation techniques to detect, normalize and enhance the image of the number plate, and then optical character recognition (OCR) to extract the alphanumeric of the license plate. ANPR systems are generally deployed in one of two basic approaches: one allows for the entire process to be performed at the lane location in real-time, and the other transmits all the images from many lanes to a remote computer location and performs the OCR process there at some later point in time. When done at the lane site, the information captured of the plate alphanumeric, date-time, lane identification, and any other information that is required is completed in somewhere around 250 milliseconds. This information, now small data packets, can easily be transmitted to some remote computer for further processing if necessary, or stored at the lane for later retrieval. In the other arrangement, there are typically large numbers of PCs used in a server farm to handle high workloads, such as those found in the London congestion charge project. Often in such systems, there is a requirement to forward images to the remote server, and this can require larger bandwidth transmission media.
27
28
6 IMAGING HARDWARE
At the front end of any ANPR system is the imaging hardware which captures the image of the license plates. License plate capture is typically performed by specialized cameras designed specifically for the task. Factors which pose difficulty for license plate imaging cameras include speed of the vehicles being recorded, varying ambient lighting conditions, headlight glare and harsh environmental conditions. Most dedicated license plate capture cameras will incorporate infrared illumination in order to solve the problems of lighting and plate reflectivity. Many countries now use license plates that are retro reflective. This returns the light back to the source and thus improves the contrast of the image. In some countries, the characters on the plate are not reflective, giving a high level of contrast with the reflective background in any lighting conditions. A camera that makes use of active infrared imaging (with a normal colour filter over the lens and an infrared illuminator next to it) benefits greatly from this as the infrared waves are reflected back from the plate. This is only possible on dedicated ANPR cameras, however, and so cameras used for other purposes must rely more heavily on the software capabilities. Further, when a full-colour image is required as well as use of the ANPR-retrieved details it is necessary to have one infrared-enabled camera and one normal (colour) camera working together. To avoid blurring it is ideal to have the shutter speed of a dedicated camera set to 1/1000 of a second. Since the car is moving, slower shutter speeds could result in an image which is too blurred to read using the OCR software, especially if the camera is much higher up than the vehicle. In slow-moving traffic, or when the camera is at a lower level and the vehicle is at an angle approaching the camera, the shutter speed does not need to be so fast. Shutter speeds of 1/500 of a second can cope with traffic moving up to 40 mph (64 km/h) and 1/250 of a second up to 5 mph (8 km/h). License plate capture cameras can now produce usable images from vehicles travelling at 120 mph (190 km/h). To maximize the chances of effective license plate capture, installers should carefully consider the positioning of the camera relative to the target capture area. Exceeding threshold angles of incidence between camera lens and license plate will greatly reduce the probability of obtaining usable images due to distortion. Manufacturers have developed tools to help eliminate errors from the physical installation of license plate capture cameras.
28
29
7 Development History
The United Kingdom: In 1992, ALPR technology was developed at Cambridge University in the United Kingdom in response to terrorism. On April 10, 1992, the Irish Republican Army bombed the Baltic Exchange, a British company operating a premier global marketplace for shipbrokers and charterers, partially destroying the faade of the Exchanges offices and extensively damaging the rest of the building (Baltic exchange, 2008). Almost a year later on April 24, 1993, the Irish Republican Army detonated a truck bomb at Bishops gate, the heart of Londons financial district (1993 Bishops gate bombing, 2008). The bombing was massive in size and destroyed several buildings with many others suffering shattered windows. The blast caused approximately 1 billion in damage, killed one person, and injured 44 others. The United Kingdom is at the forefront in the use of plate recognition technology. As a part of its intelligence network, there are 3,000 cameras across the United Kingdom, stationary (roadside) and mobile (mounted on police cars), scanning up to 5 million plates a day (Pughe, 2006). As of 2001, all 43 police forces in England and Wales were provided with vans equipped with automated number plate technology (ANPR1): ANPR cameras were mounted in vans, and police forces were provided with computers that were able to store ANPR information in real time (Pughe, 2006). The central ANPR system, i.e. the databases, is stored at the National ANPR1 This term is synonymous with ALPR and will be used interchangeably 3 Data Center (NADC) in London. NADC stores plate data and lists of suspect vehicles (Pughe, 2006). When a plate is photographed by an ANPR camera, a text file containing information on the car registration number, the time and date of the scan, and the GPS location of the camera is created. In addition, a JPEG image of the plate is produced, as well as a video image of the plate and a video of the vehicle occupants (Pughe, 2006). Approximately four seconds after a plate is read by the camera, the patrol officer is provided with information regarding the vehicle, such as whether the car was stolen, if it had been involved in another crime, or if it was uninsured (Pughe, 2006). In 2002 to 2003, the United Kingdom evaluated the use of ANPR with nine police forces. The results of this initial study indicated that officer productivity increased primarily due to the officers not having to spend as much time waiting for hits after they typed in a plate. Instead, officers were able to spend more of their time investigating hits. This resulted in an increase in arrests to 100 per year, 10 times the national average (PA Consulting Group, 2003).
29
30
In total, approximately one out of every 200 cars photographed by the ANPR cameras was stopped by an ANPR intercept team; equivalent to approximately one stop per hour. In nearly two-thirds (61 per cent) of these stops, the intercept team took some action (e.g. enforced an arrest). However, the data also highlighted the limited ability of the intercept teams to respond to the volume of hits they received. In effect, police were only able to respond to 13% of hits (PA Consulting Group, 2003). The results also indicated that the police could expect a substantial increase in the amount and value of goods they recovered. On an annual basis, on average, a constable using ANPR technology could expect to encounter: the recovery of 11 stolen vehicles (equivalent to approximately 68,000 pounds); three instances involving the recovery of other stolen goods (equivalent to approximately 23,000 pounds total); seven instances of drug seizures (equivalent to approximately 3,300 pounds total); two seizures of weapons and/or firearms; and five instances of recovery of other stolen property (PA Consulting Group, 2003). This initial study also provided the Home Office with descriptions of the quantity of vehicles on the road violating insurance or other traffic regulations, the number of vehicles used in the commission of other criminal offences, and the number of vehicles owned or operated by persons of interest (PA Consulting Group, 2003). A second evaluation was conducted between June 2003 and June 2004 with 24 police forces (PA Consulting Group, 2004). The results of this evaluation also demonstrated that the ANPR technology contributed to a substantial increase in arrests (Pughe, 2006). Over this one year period, 180,543 vehicles were stopped by intercept teams resulting in 13,499 arrests. In effect, arrests were made in nearly 4-8% of the vehicle stops. One-quarter of the arrests were for disqualified, uninsured, or prohibited drivers. A further 16.8% of arrests were made for theft or burglary offences, 10.3% for theft of or from a vehicle, and 8.2% for drug offences. In total, over 1,000 stolen vehicles were recovered over this period (PA Consulting, 2004). The results of this study were similar to the first study; ANPR was beneficial to police forces by providing better efficiency in the deployment of officers, however, workload concerns continued to exist, again suggesting a need for policy outlining a response priority scheme. Currently, ANPR technology continues to be used across the United Kingdom. For instance, the Hampshire Constabulary uses ANPR to prevent and detect terrorism, serious crime, volume crime, and fatal and serious injury road traffic accidents (Hampshire Constabulary, 2007). The Metropolitan Police Service operates four ANPR units who work across London. There are deployment teams who are used to assist in targeting hot spots for vehicle and other crime. In 2005 and 2006, these units made, an average, 200 arrests per month (Metropolitan Police Service, no date).
30
31
Charge zones the London congestion charge
The London congestion charge scheme uses 230 cameras and ANPR to help monitor vehicles in the charging zone. The London congestion charge is an example of a system that charges motorists entering a payment area. Transport for London (TfL) uses ANPR systems and charges motorists a daily fee of 10 paid before 10pm if they enter, leave or move around within the congestion charge zone between 7 a.m. and 6:00 p.m., Monday to Friday. A reduced fee of 9 is paid by vehicle owners who sign up for the automatic deduction scheme. Fines for traveling within the zone without paying the charge are 60 per infraction if paid before the deadline, doubling to 120 per infraction thereafter. There are currently 1,500 cameras, which use Automatic Number Plate Recognition (ANPR) technology. There are also a number of mobile camera units which may be deployed anywhere in the zone. It is estimated that around 98% of vehicles moving within the zone are caught on camera. The video streams are transmitted to a data centre located in central London where the ANPR software deduces the registration plate of the vehicle. A second data centre provides a backup location for image data. Both front and back number plates are being captured, on vehicles going both in and out this gives up to four chances to capture the number plates of a vehicle entering and exiting the zone. This list is then compared with a list of cars whose owners/operators have paid to enter the zone those that have not paid are fined. The registered owner of such a vehicle is looked up in a database provided by the DVLA.
The United States

Over the past several years, the United States has also introduced plate recognition technology, primarily in the form of red light cameras (Jenkins, 2007). In 2007, 150 American cities utilized this technology to catch those who drove through red lights. These offenders have a photograph taken of their plate by the ALPR cameras and receive a ticket in the mail. This technology will also be used, in several jurisdictions, to catch drivers who violate speed limits (Jenkins, 2007). The US has also used plate recognition technology at border crossings with Mexico and Canada to track auto theft (Canadian Press NewsWire, 2006). More recently, ALPR has been used to track potential terrorists. A recent study of ALPR technology in the state of Ohio emphasized
31
32
the recovery of two stolen trailers demonstrating the technologys potential utility for homeland security (McClellan, 2004). The state of Ohio studied the effectiveness of ALPR technology over a four month period in 2004, using $61,000 in federal funding. Over the four month period, beginning in August 2004, the state highway patrol apprehended 23 criminal suspects and recovered 24 stolen vehicles valued at US $220,000 (McClellan, 2004). These results were 50% higher than results during the same time period one year previous. It is important to note that the ALPR technology used in this study was only linked to databases on stolen vehicles and wanted persons and, therefore, did not provide information on unlicenced, uninsured, or prohibited drivers.
Canada
ALPR technology in Canada has only recently been implemented for use with traffic and other criminal violations. ALPR has the potential to allow Canadian police forces to engage in proactive policing. By driving through traffic corridors in cities across Canada, police can use ALPR technology linked to criminal justice and insurance databases to identify those drivers who, for instance, have outstanding warrants, are prohibited from driving, or are driving uninsured vehicles. As previously discussed, the technology also offers police a potentially useful tool for identifying and recovering stolen vehicles. ALPR technology has been used in Toronto for toll-collection purposes. The Highway 407 Express Toll Route was introduced in 1997. It allows drivers to use the express route for a toll fee that is recorded electronically. Nearly one-third (30 per cent) of the tolls are tracked through licence plate recognition technology allowing for faster movement of traffic as the need to stop and physically pay the toll is eliminated (Commission for Integrated Transport, no date). Still, there are several issues that must be considered prior to the widespread use of ALPR technology in Canada. The United Kingdom established a central data warehouse in London containing information from insurance, criminal justice, and other agencies that is linked with the ALPR technology in police cruisers. The cost of warehousing this data, in addition to the cost of buying, installing, and maintaining ALPR technology, must be weighed against the costs recovered through the use of this system, including reductions in criminal activity, recovery of stolen goods, and increases in positive public perceptions of the police (Schuurman, 2007). Additional staffing costs may also be incurred, whether in the form of additional police on the roads utilizing this technology or the hiring of additional staff to assist in confirming hits, updating the databases, and/or conducting further searches of licence plates. One way to offset these resource increases may be to, wherever possible, use civilian volunteers. ALPR was introduced in British Columbia as part of a pilot study in 2006. Schuurman (2007) recently produced an analysis of the use of this technology in parking lots in Surrey, British Columbia. Using information obtained through
32
33
the Canadian Police Information Centre (CPIC) and the Motor Vehicle Branch (MVB), lists of vehicles of interest2 were populated on a daily basis and uploaded to the patrol cars onboard computer. To collect the data for this initial study, one unmarked patrol vehicle drove through 31 Surrey parking lots scanning the licence plates of parked cars over the period of one week. Scanned plate images were compared to the CPIC and MVB databases. The Hot Lists were for unlicenced vehicles, uninsured drivers, prohibited drivers, and stolen vehicles. Over the course of this study, 21,876 licence plates were scanned. Of these, nearly all (97.7 per cent) of the scans accurately captured the cars licence plate. Schuurmans (2007) results indicated that 1.6% of scanned plates resulted in a hit. In fact, there were, on average, four hits every hour (Schuurman, 2007). Nearly three-quarters (72 per cent) of the hits occurred between 3pm and 7pm, and over two-thirds (69.8 per cent) of hits were for unlicenced drivers. The remaining hits were for unlicenced (i.e. uninsured) vehicles (23.9 per cent), prohibited drivers (4.8 per cent), and stolen cars (1.4 per cent). Schuurman (2007) concluded that the deployment of ALPR technology exclusively to parking lots would not be the most efficient use of the system, especially to locate and recover stolen vehicles. It was concluded that using the technology in a stationary format along major intersections or installing it in mobile units that patrolled the jurisdictions main roads would likely be more effective for identifying stolen vehicles. Given that there is limited information available to date on the use of ALPR technology, research must be conducted to determine the extent to which such technology is useful to Canadian police forces. Such research can be used to inform deployment strategies; for example, in determining whether it is more beneficial to have police equipped with ALPR technology drive their usual patrol routes or travel exclusively within known hot spots (Schuurman, 2007). The current study intends to provide some initial data regarding the use of ALPR technology in British Columbia, Canada.
33
34
8 Algorithm
8.1 Algorithm for ANPR system 1. Input image from webcam. 2. Convert image into binary. 3. Detect number plate area. 4. Segmentation. 5. Number identification. 6. Save to file in given format. 8.2 Input Image from file. 1. Capture image from webcam. 2. Store the captured image into a image file for further processing. 8.3 Convert image into binary. 1. Identify the intensity of the image. If image intensity = high Reduce intensity Else if intensity = low Increase intensity Else No change. 2. Convert image into grayscale. 3. Calculate appropriate threshold value for the image. 4. Convert the image into binary image using the calculated threshold. 8.4 Segmentation 1. Crop the image. 2. Filter the noise level present in the image. 3. Clip the plate area in such a way that only numbers of plate area extracted. 4. Separate each character from the plate. 8.5 Number identification 1. Create the template file from the stored template images. 2. Resize image obtained from segmentation to the size of template.
34
35
3. Compare each character with the templates. 4. Store the best matched character. 8.6 Save to file in given format 1. Open a text file in write mode. 2. Store the character obtained from the number identification process to text file in given format. 3. Close the file.
35
36
Overview of the Vehicle license plate System
Fig 8.1
36
37
9 Image Acquisition The initial phase of image processing for Vehicle License Plate Recognition is to obtain images of vehicles. Electronic devices such as optical (digital/video) camera, webcam etc can be used to capture the acquired images. For this project, vehicle images will be taken with a Panasonic FX 30 digital camera. The images will be stored as colour JPEG format on the camera. Next, we might proceed in using the Matlab function to convert the vehicle JPEG image into gray scale format. 9.1 Image Cropping Image cropping is a recognition process whereby it will extract the smallest rectangle which will contains the edge of the license plate and license plate itself. As the license plate surrounding is of no importance, this cropping process will highly increase the speed of image processing. The figure below shows the copping of image:
Fig 9.1: Original Image
Fig 9.2 : Cropped Figure

37
38
9.2 Conversion of RGB Images to Binary Images In order facilitate the next process smoothly and reducing the processing time, the image obtain is being converted to Binary Image. The conversion is done by using the toolbox in MATLAB (im2bw). 9.2.1 What is Binary Images It is an image which quantatised into two values representing 0 and 1 or in pixel values of 0 and 255 representing the colour black and white. Binary images is the simplest process and has apply to many other application. It is useful as the information we need can be obtained from the silhouette of the object. The application is as: Text interpreting Identify the object orientations Next, Binary Images are obtained by converting the input image into grayscale format, then by converting the grayscale image to binary image by thresholding. The image is made up of a matrix squares which is called Pixel. Each pixel in the image has a brightness value which is known as grey level. The pixel of grey level above the threshold will be set to 1 (equal to 255; white) and the rest will be set to 0 (black). We will obtain white object with black background or vice versa. The characteristic function for the image object is classify below:
Fig 9.3
38
39
Below will illustrate different type of histogram images Fig 9.4: Histogram of light object with dark background. Frequency
Fig 9.5 : Histogram of object and background which is close
39
40
In this circumstance, there is difficulty in choosing the threshold as the histogram is no more bimodal This root cause is due to the variations of light or colours in the images. 9.3 Colour Image Processing RGB Colour Space The RGB image is made of colour pixels of an M x N x 3 array. The colour space is normally graphically shown as RGB colour cube. The cube vertex consists of the primary colour (Red, Green and Blue) and the secondary colour (Cyan, Magenta and Yellow).The schematic of the RGB cube that illustrate both the primary and secondary colour at each vertex is shown below.
Fig 9.6
40
41
Fig 9.7 : Original Image
Fig 9.8 : Binary mage
41
42
10 IMAGE ENHANCEMENT The aim of this process is to increase and improve the visibility of the image. Image Enhancement techniques consists process of sharpening the edges image, contrast manipulation, reducing noise, color image processing and image segmentation as well.
10.1 IMCOMPLEMENT
Imcomplement will compute the complement image(IM). IM can intensity, binary or true color image. For the complement binary image, black will becomes white and white will be come black ; zeros will become ones or ones will become zeros. For the complement of the true color image, the value of each pixel will be subtracted by the maximum pixel value which is supported by the class. The pixel value obtained by the differences will be used in the output image. Therefore, the light areas will become darker and the darker area will become lighter.
10.2 MEDFILT2
Median filtering is a nonlinear operation often used in image processing to reduce "salt and pepper" noise. A median filter is more effective than convolution when the goal is to simultaneously reduce noise and preserve edges.
42
43
11 Thresholding Image thresholding enjoys the central position of any application in the process of image segmentation. One of the simplest process will be the Gray-level thresholding. The regions of the images are classified by the reflectivity and absorption of light on its surface. Thresholding is one of the oldest segmentation method. It is still greatly used in applying on the applications in the market due to certain reason; Fast and cheap. The value of threshold (T) is being selected and compare with the pixel of the image. It also transform the input image(K) into an output binary image (F) which is being segmented. F (x,y) = 1 if K (x,y)>=T; = 0 if K (x,y) < T; Representing F (x,y) =1 for image object; F (x,y) = 0 for background of the object and T = threshold.
11.1 Threshold Selection

The key parameter in the thresholding process is the choice of the threshold value (or values, as mentioned earlier). Several different methods for choosing a threshold exist; users can manually choose a threshold value, or a thresholding algorithm can compute a value automatically, which is known as automatic thresholding . A simple method would be to choose the mean or median value, the rationale being that if the object pixels are brighter than the background, they should also be brighter than the average. In a noiseless image with uniform background and object values, the mean or median will work well as the threshold, however, this will generally not be the case. A more sophisticated approach might be to create a histogram of the image pixel intensities and use the valley point as the threshold. The histogram approach assumes that there is some average values for both the background and object pixels, but that the actual pixel values have some variation around these average values. However, this may be computationally expensive, and image histograms may not have clearly defined valley points, often making the selection of an accurate threshold difficult. In such cases a unimodal threshold selection algorithm may be more appropriate. One method that is relatively simple, does not require much specific knowledge of the image, and is robust against image noise, is the following iterative method: 1. An initial threshold (T) is chosen, this can be done randomly or according to any other method desired.
43
44
2. The image is segmented into object and background pixels as described above, creating two sets: 1. G1= {f(m,n):f(m,n)>T} (object pixels) 2. G2= {f(m,n):f(m,n) T} (background pixels) (note, f(m,n) is the value of the pixel located in the mth column, nth row) 3. The average of each set is computed. 1. m1= average value of G1 2. m2= average value of G2 4. A new threshold is created that is the average of m1and m2 1. T = (m1+m2)/2 5. Go back to step two, now using the new threshold computed in step four, keep repeating until the new threshold matches the one before it (i.e. until convergence has been reached). This iterative algorithm is a special one-dimensional case of the k-means clustering algorithm, which has been proven to converge at a local minimum meaning that a different initial threshold may give a different final result.
11.2 Global Thresholding This is a method whereby the histogram of the image is being partitioned using a single threshold value. Image is form up of square matrixes called pixels and each pixels in the image have a brightness level known as gray level. The value of the gray level varies from the darkest (0) to the brightest (255). The threshold defines the value of the gray level in which the baseline boundary is in between the foreground pixels and the background. We can use the graythresh toolbox function to calculate the threshold value by using Otsus method. The threshold value which is selected is use to reduce the intraclass variance between the white and black pixels.
11.3 Adaptive Thresholding

The vehicle number plate can be partially illustrated. This is a common reason why global thresholding fail. By using the adaptive thresholding, its can solve the disadvantages of the global thresholding as it computes the value of the threshold for every pixel separately using the neighborhood. If the histogram is bimodal, the value of the threshold will be taken in the middle of the bimodal. For other cases, in which the modes are overlapping,, the threshold will try to reduce the error of taking the background pixels as object pixel and vice versa.
44
45
12 IMAGE NOISE Digital camera with long exposure time setting and high ISOs will produced images with lots of noise. Noise in the images is appears as the colour speckles whereby there should not be any. The reason of noise contamination on the image is categorised as below: There will be not enough light source for proper exposure to activate in a low light area/ Therefore, we need longer time for the image sensor for collection of weak signal. In this circumstance, the electrical noise from the background will be higher than the signal. We will be enlarging the signal from the received light photons when we are using the digital camera with a higher ISO setting. From the amplification of the signal, the electrical noise in the background is also being amplified.
12.1 Filtering
In this pre-processing stage, filtering process is used for blurring and for noise reduction and thus resulting a more sharpen image. Blurring is used in the preprocessing step which involved in removing and cleaning up of small patches and details from the image prior to object extraction, bridging of gaps in curves and lines. Next, blurring with liner/non liner filter is able to achieve noise reduction. We can apply convention techniques such as softening and sharpening. In order to reduce the background noise, softening is often applied to the low pass filter and it will affect the image to become blurring.
12.2 Binary area open (Bwareaopen) The MATLAB toolbox function provide a bwareaopen function in which it removes the connected component which have fewer than P pixel. Thus, it will reproduce another binary image. The syntax is as bellowed: BW2 = bwareaopen(BW,P);
12.3 FILTERING OF DIGITS

Now, we have managed to obtain the black and white image of the Vehicle License Plate. It has been cropped to a rectangular shape in which it enhances the algorithm to be more uniformed in the further stages.
45
46
The non-digits substance which appears in the image is caused by the noise (quantization that will transform to Black), object or substance which is not digits/characters such as screws, sand, water droplet, frame lines, effects on JPEG compression. By Filtering, we are able to remove or filter out the unwanted substances or noise that is not a character or digits. Lastly, the image is only left with characters and digits in which we are interested. The 2 stages for the algorithm is as follows: 1) Remove out the small objects or connected components. 2) Identify the frame line that is connected to the digits and separate it. 12.4 Identify and remove away the small connected objects For this stage, we need to identify the connected components. The component can have a value of either 4-connected or 8-connected. In this algorithm, we use 8-connected (8 specify 8-connected objects). After the component has been labelled, they will obtain a unique number; The elements of labels (stated in the algorithm) contains integer values that will be greater than or equal to 0. The pixel of 0 represents the background and pixel of 1 will make up the first object, the pixel of 2 will label the second object, so on and so for. Below illustrate an example of filtering process.
Fig 12.1 : Binary image with Noise For the above image, 141 connected components are found in the BW image.
46
47
Fig 12.2 : Binary Image after Filtering. Next, we have apply the median filter and bwareaopen (Image Processing Toolbox) whereby it will remove all the connected components from the binary image that have value less than P pixels, in which it will produce another binary image. Therefore, components with less than 6000 is classify as non digits components and need to be taken away.
47
48
13 Image Segmentation
Image segmentation plays an important and critical step that lead to the analysis of the processed image data. In order to extract and analyzed the object characteristic, the process need to partition the image into different parts that will have a strong correlation with the objects. Segmentation process can be categorized into several parts. Firstly is the global Knowledge of an image. The feature of the image is represented by a histogram. Secondly is the boundary-based segmentation. The process uses the edge detection to obtain the region contours and the objects will construct from the obtain contours. Lastly will be the edge-based segmentation. For this project, we will be trying out first category segmentation method, which is by using threshold.
13.1 FEATURE EXTRACTION FROM DIGITAL IMAGE

The digital image description is depends on the external and internal representation. The color or texture of the image is basically the internal representation whereas the external representation is based on the characteristic of the shapes. The normalized character description is based on the external characteristics as we only work for properties on the shape of the character. The descriptor vector includes the characteristics as the number of lines, vertical or diagonal edges etc. The process of the feature extraction is to transform the bitmap data into a form of descriptor in which more suitable for computer. If we classify the similar character into classes, the descriptor of the character from the same class is close to each other in the vector space. This will lead to a success in pattern recognition process. The algorithm below will represent how the extraction is being carried out and extracted figure is illustrated as below. [L Ne]=bwlabel(imagen); disp(Ne); for n=1:Ne [r,c] = find(L==n); n1=imagen(min(r):max(r),min(c):max(c)); imshow(~n1); img_r=same_dim(~n1); letter=read_letter(img_r); word=[word letter]; end
48
49
Plotted Bounding Box
Fig 13.1
13.2 Database
It is a collection of information or data which it is being orderly organize, thus it can be accessed easily and updated. Database can be in the form of text, contents and images. Database is needed to make sure that the image space can contained enough characters which have been extracted and the vehicle license plated number stored in the excel sheet for the purpose of comparison. The database would be enlarged in order to improve the accuracy and better chances of obtaining the correct result. The class recognition will check if the computed ratio is correct. It will ignore the class if the image thresholding is too low. If the class has match correctly, it will be compare with data set in the database and provide an output if the image is recognized correctly.
49
50
14 Character Recognition 14.1 Normalization

In this phase, the extracted characters are resized to fit the characters into a window. For the project, each character is normalized to the size of (42x24) binary image and then follow by reshape to standard dimension before sending the data set to the next for comparison with the templates.
14.2 Template Matching

The correlation between two signals (cross correlation) is a standard approach to feature detection as well as a component of more sophisticated techniques. Textbook presentations of correlation describe the convolution theorem and the attendant possibility of efficiently computing correlation in the frequency domain using the fast Fourier transform. Unfortunately the normalized form of correlation (correlation coefficient) preferred in template matching does not have a correspondingly simple and efficient frequency domain expression. For this reason normalized cross-correlation has been computed in the spatial domain. Due to the computational cost of spatial domain convolution, several inexact but fast spatial domain matching methods have also been developed. An algorithm for obtaining normalized cross correlation from transform domain convolution has been developed, see Lewis . The new algorithm in some cases provides an order of magnitude speedup over spatial domain computation of normalized cross correlation. 14.2.1 Cross Correlation The use of cross-correlation for template matching is motivated by the distance measure (squared Euclidean distance) .
(where f is the image and the sum is over x,y under the window containing the feature t positioned at u,v). In the expansion of d2
50
51
the term is constant. If the term constant then the remaining cross-correlation term
is approximately
is a measure of the similarity between the image and the feature.
51
52
15 ANALYSIS OF RESULT
Tests are conducted according to the modular form
15.1 Extracting of Individual Digits

Determine the angle of the Vehicle License Plate. It is generally important when capturing the vehicle image. For instance, the figure illustrate below lead to wrong recognition of the character.
Fig 15.1 Recognition : D6LH4CNEB800
Fig 15.2 : Recognition: MH01AX7946

52
53
To improve in the cropping of image, we can improve on the accuracy of capturing of vehicle image. Next, we can also include an algorithm where we can change the image of the four coordinates to a standard rectangle size or manually crop the image.
15.2 Vehicle Quantization and Equalization

For some rare cases, such as very dark image, high contrast image, low contrast image, the binarized image do not allow to make the difference between the background and the digits. This will result fail recognition of the vehicle license plate. To improve on the performance of the character recognition, we can make the difference between the digits and background inside the license plate. Equalization and quantization allow to obtain a gray scale image with improve contrast between digit and the background.
15.3 Checking and Verification of the Statistic

In the midst of trying out other car plate, I have experience loosing out character when performing the extracting process. After much study the MATLAB toolbox syntax, we have managed to make adjustment on the value and apply the skill of binary area open (BWAREAOPEN) in which help to remove the small object too. Below illustrate a example of the scenario: Morphologically open binary image (remove small objects) imagen = bwareaopen(imagen,6000); figure,imshow(imagen);
Fig 15.3
53
54
Morphologically open binary image (remove small objects) imagen = bwareaopen(imagen,7000); figure,imshow(imagen);
Fig 15.4
54
55
16 List of functions used

1. Imread : Read image from graphics file. 2. Rgb2gray - Convert RGB image or colormap to grayscale. 3. Graythresh - Global image threshold using Otsu's method. 4. Im2bw - Convert image to binary image, based on threshold. The output image BW replaces all pixels in the input image with luminance greater than level with the value 1 (white) and replaces all other pixels with the value 0 (black). 5. Size - returns the sizes of each dimension of an image in matrix a and b. 6. Imcrop - creates an interactive Crop Image tool associated with the image displayed in the current figure, called the target image. The Crop Image tool is a moveable, resizable rectangle that you can position interactively using the mouse. When the Crop Image tool is active, the pointer changes to cross hairs when you move it over the target image. Using the mouse, you specify the crop rectangle by clicking and dragging the mouse. You can move or resize the crop rectangle using the mouse. When you are finished sizing and positioning the crop rectangle, create the cropped image by double-clicking the left mouse button or by choosing Crop Image from the context menu. imcrop returns the cropped image 7. Imcomplement - computes the complement of the image. In the complement of a binary image, zeros become ones and ones become zeros; black and white are reversed. 8. Medfilt2 - Performs median filtering of the matrix A using the default 3by-3 neighborhood. 9. Fopen - Open file, or obtain information about open files. 10.Bwlabel - returns in num the number of connected objects found in BW. 11.Bwareaopen - Morphologically open binary image (remove small objects). 12.Regionprops - measures a set of properties for each labelled region.
55
56
13.Rectangle - draws a rectangle with Position [0,0,1,1] and Curvature [0,0] (i.e., no curvature). 14.Disp - displays an array, without printing the array name. 15.Max- Largest elements in array. 16.Min - Smallest elements in array. 17.Find - Find indices and values of nonzero elements. 18.Resize - Resize image. 19.Corr2 - 2-D correlation coefficient. 20.Fprintf - Write text to device. 21.Isempty - Determine whether array is empty. 22.Fclose - Close one or more open files. 23.Imshow - Display image.
56
57
17 CODE 17.1 Algorithm to extract characters ( extractdigit.m).

image=imread('exp4.jpg'); % Show image figure,imshow(image); title('INPUT IMAGE WITH NOISE') % Convert to gray scale if size(image,3)==3 % RGB image imagen=rgb2gray(image); end % Convert to binary image threshold = graythresh(image); image =~im2bw(image,threshold); figure,imshow(image);title('binary image');%pause(10); image = medfilt2(image); % Remove all object containing fewer than 6000 pixels image = bwareaopen(image,6000); pause(1) image = imcrop(image); % Show image binary image figure,imshow(~image); title('INPUT IMAGE WITHOUT NOISE') word=[];%Storage matrix word from image fid = fopen('log.txt', 'at');%Opens a text for append in order to store the number plates for log. [L num]=bwlabel(image); disp(num); propied=regionprops(L,'BoundingBox'); hold on for n=1:size(propied,1) rectangle('Position',propied(n).BoundingBox,'EdgeColor','g','LineWidth',2) end hold off pause (1)
57
58
figure for n=1:num [r,c] = find(L==n); n1=image(min(r):max(r),min(c):max(c)); figure,imshow(~n1); img_r=same_dim(~n1); letter=read_letter(img_r); word=[word letter]; pause(0.5) end fprintf(fid,'Number Plate:-%s\nDate:-%s\n',word,date); fprintf(fid,'------------------------------------\n'); msgbox(sprintf('Number Plate Extraction successful.\nExtracted Number plate:- %s .\nSee the log.txt file to see the stored number.',word),'Extraction Success'); word=[]; fclose(fid);
17.2 Algorithm of Normalisation of characters (same_dim.m)

function img_r=same_dim(imagen_g) img_r=imresize(imagen_g,[42 24]);
17.3 Algorithm to read letters (read_letter.m)

function letter=read_letter(imagn) comp=[]; load templates for n=1:36 sem=corr2(templates{1,n},imagn); comp=[comp sem]; end vd=find(comp==max(comp)); if vd==1 letter='A'; elseif vd==2 letter='B'; elseif vd==3
58
59
letter='C'; elseif vd==4 letter='D'; elseif vd==5 letter='E'; elseif vd==6 letter='F'; elseif vd==7 letter='G'; elseif vd==8 letter='H'; elseif vd==9 letter='I'; elseif vd==10 letter='J'; elseif vd==11 letter='K'; elseif vd==12 letter='L'; elseif vd==13 letter='M'; elseif vd==14 letter='N'; elseif vd==15 letter='O'; elseif vd==16 letter='P'; elseif vd==17 letter='Q'; elseif vd==18 letter='R'; elseif vd==19 letter='S'; elseif vd==20 letter='T'; elseif vd==21 letter='U'; elseif vd==22 letter='V'; elseif vd==23 letter='W'; elseif vd==24 letter='X';
59
60
elseif vd==25 letter='Y'; elseif vd==26 letter='Z'; %*-*-*-*-* elseif vd==27 letter='1'; elseif vd==28 letter='2'; elseif vd==29 letter='3'; elseif vd==30 letter='4'; elseif vd==31 letter='5'; elseif vd==32 letter='6'; elseif vd==33 letter='7'; elseif vd==34 letter='8'; elseif vd==35 letter='9'; else letter='0'; end
60
61
18 Result Analysis and Discussion 18.1 Experimental results
Extracted Number plate: MH12EM2587
Extracted Number plate: MH12FF7461
Extracted Number plate: MH12DM3909
61
62
Extracted Number plate:LMH12CD9736
Extracted Number plate:LK00A53N6523
Extracted Number plate: MH12HN0389
62
63
Extracted Number plate: MH14CS7710
Extracted Number plate:MH12DE94
Extracted Number plate: MH01AX7946
18.2 Discussion
Although ALPR technology has been promoted as being capable of reading in excess of 3,000 plates per hour, the current results suggest that this capability is significantly overstated in actual road tests. The technology may have the capacity to read more plates, but traffic volume and the design of Surrey streets may impede this read rate. Perhaps the most important conclusion reached from this study is that the utility of ALPR depends on volume. The initial analysis of parking lot data produced by Schuurman (2007) suggested that parking lot
63
64
deployment was dependent upon the number of vehicles in parking lots and, for the most part, the same conclusions applied to this road test of the technology. In both cases, the more cars scanned, the greater the number of raw hits. Importantly, the nature of hits was basically uniform for all of the assigned traffic corridors and the proportions held when considered by time of day or day of month. In effect, it was all about the number of hits, and, for the most part, the frequency of hits exceeded what a typical patrol unit could respond to during a shift. As a consequence, the fact that officers could expect several hits per hour requires the design of a response priority scheme and increase patrol units to manage the increased workload. However, considering the results of this phase of the project, in order to maximize efficiency, police forces operating ALPR technology may want to focus on high volume traffic corridors during the day shift. Given this, it may be possible, as suggested above, to train volunteers who can assist the police in filtering through database hits, substantiating which calls are valid and which must receive priority attention to lessen the burden on patrol officers. Unfortunately, these statistics were not improved in the current analysis. Despite the promise that ALPR technology holds for catching stolen vehicles, the current study results also indicated that stolen vehicles represented a very small proportion (approximately one per cent) of all licence plate hits. However, this result should not have been unexpected. There are two main reasons why ALPR, as currently operated in Surrey, was not effective in locating stolen vehicles. First, the hit list that the ALPR-enabled vehicle searched against was 24 hours old. Second, ALPR requires a meeting in time and space between the police cruiser and the stolen vehicle. In other words, for a stolen vehicle to be detected by an ALPR-enabled vehicle, the stolen car had to be on the road in Surrey, along one of the four traffic corridors assigned for that day, at least 24 hours after the car had been stolen, and at the exact location where a moving ALPR-enabled vehicle could photograph its licence plate. The likelihood of this scenario occurring is extremely low. Rather, until such time as ALPR-enabled vehicles have access to real-time data and there are many more ALPR-enabled police cruisers on the road at the same time, it may be more beneficial to use stationary cameras along a large number of intersections, if the objective is to identify stolen vehicles as this will, at least, remove the variable of a roving ALPR which likely reduces success. An important consideration raised by this research has to do with prioritizing responses and the allocation of scarce resources. If patrol officers are faced with a significant increase in the number of hits as a result ALPR deployment, detachments must design response schemes to prioritize which hits officers respond to. A large advertising campaign can inform the public about the nature and use of this technology which may result in the public being less willing to violate traffic regulations.
64
65
19 Difficulties and Recommendations

There are a number of possible difficulties that the software must be able to cope with. These include:

Poor image resolution, usually because the plate is too far away but sometimes resulting from the use of a low-quality camera. Blurry images, particularly motion blur. Poor lighting and low contrast due to overexposure, reflection or shadows. An object obscuring (part of) the plate, quite often a tow bar, or dirt on the plate. A different font, popular for vanity plates (some countries do not allow such plates, eliminating the problem). Circumvention techniques. Lack of coordination between countries or states. Two cars from different countries or states can have the same number but different design of the plate.
While some of these problems can be corrected within the software, it is primarily left to the hardware side of the system to work out solutions to these difficulties. Increasing the height of the camera may avoid problems with objects (such as other vehicles) obscuring the plate but introduces and increases other problems, such as the adjusting for the increased skew of the plate. On some cars, tow bars may obscure one or two characters of the license plate. Bikes on bike racks can also obscure the number plate, though in some countries and jurisdictions, such as Victoria, Australia, "bike plates" are supposed to be fitted. Some small-scale systems allow for some errors in the license plate. When used for giving specific vehicles access to a barricaded area, the decision may be made to have an acceptable error rate of one character. This is because the likelihood of an unauthorized car having such a similar license plate is seen as quite small. However, this level of inaccuracy would not be acceptable in most applications of an ANPR system. When running the main.m file, it is able to detect and recognize some of the zoom in car vehicle license plate but fail on the image of cars with headlights. For the future works and suggestion on improvements, these are the steps which are recommended: Modification is needed to be done on the offset of detecting the rectangular plate or by applying other technique to the system.
65
66
We only use the black and white image of the Vehicle license plate. For future implementation, we can use RGB to HSV or RGB to CMY method to deal with other type of colour license plate. There should be improvement on the decision of the algorithm and ways to detect error. When the probability of recognition guess is correct but falls below the threshold, the recognition system should refuse to make the decision.
66
67
20 Critical Review and Reflection

The development of the project is considered quite successful although we still faced some problems during the progress of the project we managed to solve the problem by doing more reading, research and with the help and understanding from the tutor. The problem encountered during the progress of the project include the selection of suitable method such as : sourcing of car images, methods on detection, method of segmentation and recognition of vehicle license plate character and writing of simulation program. During the initial project and development stage, the problem encountered is the selection of software program. After doing much reading and research, we chose MATLAB software as it is easier to implement. Next, we also encountered problems in detection, segmentation, recognition of vehicle license plate characters. After doing much research and absorbing more knowledge into MATLAB toolbox, problems were resolved by using the suitable syntax and methods.
67
68
21 Conclusion & Future Scope

In this report, we presented application software designed for the recognition of car license plate. Firstly we extracted the plate location, then we separated the plate characters individually by segmentation and finally applied template matching with the use of correlation for recognition of plate characters. This system is designed for the identification INDIAN license plates and the system is tested over a large number of images. Finally it is proved to be %96 for the segmentation of the characters and %80 for the recognition unit accurate, giving the overall system performance 92.57% recognition rates. This system can be redesigned for multinational car license plates in future studies. The ALPR technology offers several benefits to police forces. Most importantly, ALPR has the ability to quickly and efficiently scan a large number of licence plates without any officer intervention, such as having an officer physically type in a licence plate to scan. ALPR also offers an objectivity that may benefit police. Given that officers cannot check all the licence plates they encounter while on shift, they are compelled to make a series of decisions concerning which plates to search. While officers currently employ a set of indicators, identified through experience, with which to select those plates that appear more suspicious, it is highly plausible that through this process, officers may miss plates that are, in fact, untoward. The ability of ALPR to scan a large number of plates allows for more plates to be scanned faster and more efficiently. The use of ALPR technology might also result in safer police driving as officers would no longer have the added distraction of turning away from the road periodically to manually type in licence plates of interest . Research with ALPR has shown several benefits, namely, increased police efficiency. With an increased number of hits, or successful matching between a scanned plate image and a database of interest, police are better able to identify more persons of interest. This increases the potential for the recovery of stolen goods as well as convictions. The technology also allows the police to identify uninsured vehicles, prohibited drivers, and unlicensed drivers much more quickly than previous police strategies. There are, however, some limitations inherent in the use of such advanced technology. With the potential for an increased number of hits, officers could become overwhelmed by the increase in the number of problematic cars to respond to. Research in the United Kingdom suggested that in responding to the sheer number of hits identified through ALPR, an officers workload substantially increased, impeding their ability to efficiently respond not only to ALPR hits, but to other calls for service. As a consequence, police must develop strategies that enable officers to prioritize their responses. However, as the profile of crime is different in jurisdictions, detachment-specific schemes may need to be developed. In other
68
69
words, depending on the geographic location of hotspots, the number of officers on patrol, and the specific needs of the community, priority schemes may need to be individualized. Moreover, to better respond to priority hits, police forces may find it necessary to increase the number of officers on the road which, given current fiscal realities in many detachments in Canada, may not be feasible. While advances in technology allow for the successful use of systems, such as licence plate recognition, it also provides new methods with which to avoid being screened by such technology. For instance, Gordon and Wolf (2007) reported that since the advent of ALPR, some companies have begun to sell products to thwart the technology. They noted that one company sells a clear spray (US $30 per can) that the manufacturers claim can make licence plates invisible when read by a camera. In effect, as the police develop new technologies, there will be those who develop the means to defeat these techniques. In the past, recognition software has produced extremely low successful recognition rates (Gordon and Wolf, 2007). Yet, more recently, research suggested that the ALPR technology reads plates correctly 95% of the time. However, in the event that a plate is not read correctly and the officer deems the plate suspicious, it is important that officers to retain the ability to investigate the plate more fully and to follow their instincts when observing a suspicious vehicle. The efficiency of ALPR technology is entirely dependent upon the successful coordination of agencies. Without the provision of data by which to compare scanned plates, ALPR cannot possibly identify plates of interest. Information can either be provided in real time, e.g. lists are updated as cars or plates are reported stolen or it can be updated every 24 hours. Either way, those using ALPR technology must have access to data that details information about stolen plates or cars, vehicles that have been involved in other criminal activity, drivers that have been prohibited or have lost their licence, or drivers who are uninsured. Much of the work involved in implementing ALPR technology involves building these initial relationships between agencies. Again, without the successful coordination between police forces and agencies, such as insurance companies, other criminal justice agencies, and the motor vehicle branch, ALPR technology simply will not succeed. Lastly, privacy concerns are also a limitation to the use of ALPR technology. Concerned citizens may accuse police or the government of using the technology to track law-abiding citizens, invading their right to privacy. Citizens may equate the use of ALPR technology to fishing expeditions, where police simply scan all plates until they get a hit, as opposed to specifically searching out particular plates based on prior intelligence. Concerns in Canada already exist regarding the use of Closed Circuit Television (CCTV) systems in public (Schuurman, 2007). Deisman (2003) identified that there are limits to the extent that police in Canada can engage in continuous and non-selective monitoring of citizens. Schuurman
69
70
(2007) also noted that the Canadian Charter of Rights and Freedoms states that privacy rights of citizens are breached by indiscriminate video surveillance without cause. Essentially, more research needs to be conducted in order to determine how ALPR technology can be balanced with respect to citizens right for privacy and civil rights. Citizens may also have concerns with respect to the maintenance of data in warehouses. Citizens may fear the potential for breaches in security. In addition, there may be concerns regarding who has access to this data . It is, therefore, extremely important that considerable thought is given to the safe storage of data and strict regulations regarding who has access to the databases. In responding to concerns of privacy, policies may be put in place that regulate the deleting of collected data on a daily, weekly, monthly, or yearly basis. In conclusion, although further research is needed to determine the extent to which ALPR increases the rate of arrest and has a deterrent effect, the results of this study suggest that ALPR technology offers several substantial benefits to the police. However, it is difficult to assess the impact of ALPR on police resources and workloads, and thus make any firm conclusions on its general utility. Still, ALPR does have a specific utility as its strategic deployment will assist police departments to more effectively response to a variety of auto and driving-related offences. The overriding benefit of ALPR is that it brings a far larger number of offenders to the attention of the police, rather than the few offenders the police are able to find during their routine activities. How to adequately respond to this situation will require careful thinking and planning on the part of the police.
Future Scope
Though we have achieved an accuracy of 80% by optimizing various parameters, it is required that for the task as sensitive as tracking stolen vehicles and monitoring vehicles for homeland security an accuracy of 100% cannot be compromised with. Therefore to achieve this, further optimization is required. Also, the issues like stains, smudges, blurred regions & different font style and sizes are need to be taken care of. This work can be further extended to minimize the errors due to them.
70
71
Annexure A: Project Plan & Progress Report

For developing the project, we have drafted out the project schedule to follow up closely. Task1 Research (24 Jan 2012 to 10 Feb 2012) Research on various methods used for Vehicle License Plate Recognition Conversion of Images Image Thresholding Noise filtering Vehicle License Plate Detection Vehicle License Plate Extraction/Segmentation Vehicle License Plate Recognition
Task 2 Development (11 Feb 2012 to 23 Feb 2012) Determine the value of pixel need for filtering Determine algorithm for plate detection Determine algorithm for extraction/segmentation Determine algorithm for plate recognition. There is interconnected link between the research work and the designing and Implementation process. Thus, it is important to choose the type of model and syntax that are suitable for Matlab is taken into consideration.
Task 3 Construct Matlab software for simulation (24 Feb 2012 to 20 Mar 2012) Writing and create algorithm for extraction of vehicle license plate character Writing and simulate the block by integrating all the written algorithm together Version R2008a for editing.
Task 4 Testing and Fine Tuning (30 March 2012 to 24th April 2012)
Fine tune on the filtering and calculation of connected components. Fine tuning of the integrated block algorithm Task 5 Reporting Writing (28 April 2008 to 1 May 2012) Allocate at least 1 week for writing the final report.
71
72
References
Rafael C. Gonzalez, Richard E. Woods and Steven L. Eddins, Digital ImageProcessing using MATLAB Rafael C. Gonzalez and Richard E. Woods, International Edition Digital Image Processing Second Edition John C. Russ, The Image Processing Handbook Adrian A.Hopgood,Intelligent Systems for Engineers and Scientists William J. Palm III, Introduction To Matlab 7 For Engineers Magrab,Edward B,An Engineers Guide To Matlab ALGORITHMIC AND MATHEMATICAL PRINCIPLES OF AUTOMATIC NUMBER PLATE RECOGNITION SYSTEMS. Author ONDREJ MRTINSKY Lecture notes Mathworks.com/access/helpdest/help/techdoc/mathlab_product_page2.ht ml#printable_pdf. Mathworks.com/access/helpdesk/help/toolbox/nnet/nnet_product_page.ht ml Matlab software version R2008a. It is a very user friendly device as it will guide and explain to us on the specific toolbox in which we will be using. CYNTHIA LUM, LINDA MEROLA, JULIE WILLIS, AND BREANNE CAVE , Center for Evidence-Based Crime Policy, George Mason University , LICENSE PLATE RECOGNITION TECHNOLOGY (LPR) IMPACT EVALUATION AND COMMUNITY ASSESSMENT . Dr. Irwin M. Cohen, Dr. Darryl Plecas, and Amanda V. McCormick SCHOOL of CRIMINOLOGY and CRIMINAL JUSTICE University College of the Fraser Valley, A Report on the Utility of the Automated Licence Plate Recognition System in British Columbia
72
73
D.G. Bailey, D. Irecki, B.K. Lim and L. Yang Test bed for number plate recognition applications, Proceedings of First IEEE International Workshop on Electronic design, Test and Applications ( DELTA02 ), IEEE Computer Society, 2002. R.A. Lotufo, A.D. Morgan, and AS. Johnson, 1990, Automatic NumberPlate Recognition, Proceedings of the IEE Colloquium on Image analysis for Transport Applications, V01.035, pp.6/1-6/6, February 16, 1990. A.S. Johnson, B.M. Bird, 1990, Number-plateMatching for Automatic Vehicle Identification, IEE Colloquium on Electronic Image and Image Processing in Security and Forensic, Aprl, 1990. M.M.M. Fahmy, 1994, Automatic Number-plate Recognition : Neural Network Approach, Proceedings of VNIS94 Vehicle Navigation and Information System Conference, 3 1 Aug-2 Sept, 1994 J.A.G. Nijhuis, M.H. Ter Brugge, K.A. Helmholt, J.P.W. Pluim, L. Spaanenburg, R.S. Venema, M.A. Westenberg, 1995, Car License Plate Automatic Vehicle Identification by Plate Recognition Author - Serkan Ozbay, and Ergun Ercelebi. http://en.wikipedia.org/wiki/Vehicle_registration_plates_of_India http://en.wikipedia.org/wiki/Automatic_number_plate_recognition http://www.anpr.net/ http://www.platerecognition.info/
73
74
74

Mayank Finalproject

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mayank Finalproject

Uploaded by

Copyright:

Available Formats

1

LICENSE PLATE RECOGNITION SOFTWARE

ABHINANDAN SINGH - 08070121202 MAYANK KISHORE PATHIKRIT GUHA - 08070121224 - 08070121232

Under the Guidance of

Asst. Prof. MRUDANG SHUKLA

Prof. Mrudang Shukla

Prof. Dr. Narayan Pisharoty

Prof. Mrudang Shukla

Prof. Dr. Narayan Pisharoty

Prof. Mrudang Shukla

Prof. Dr. Narayan Pisharoty

The Project entitled

LICENSE PLATE RECOGNITION SOFTWARE

Abhinandan Singh - 08070121202 Mayank Kishore - 08070121224 Pathikrit Guha - 08070121232

template matching Median Filter Connected Components Masking.

2 Problem Definition 2.1 Problem statement:

3.2 How the Vehicle License Plate System works

Fig 3.2 A car approaching a License Plate Recognition System

System block Diagram

This scheme of numbering has some advantages:

3.6 Commercial Products

3.6.1 IMPS (Integrated Multi-Pass System)

3.6.4 Hi-Tech Solution

Format Name TIFF JPEG GIF BMP PNG XWD

Extension .tif; .tiff .Jpg ; .Jpeg .gif .bmp .png .xwd

4.2 Image Prcessing

Block Diagram of Image Processing Steps

4.3 Comparison Between MATLAB and Artificial Intelligence

Advantage of Artificial Intelligence:

Disadvantage of Artificial Intelligence:

Charge zones the London congestion charge

The United States

Overview of the Vehicle license plate System

Fig 9.1: Original Image

Fig 9.2 : Cropped Figure

Fig 9.5 : Histogram of object and background which is close

Fig 9.7 : Original Image

Fig 9.8 : Binary mage

11.1 Threshold Selection

11.3 Adaptive Thresholding

12.3 FILTERING OF DIGITS

13.1 FEATURE EXTRACTION FROM DIGITAL IMAGE

Plotted Bounding Box

14 Character Recognition 14.1 Normalization

14.2 Template Matching

is a measure of the similarity between the image and the feature.

15.1 Extracting of Individual Digits

Fig 15.1 Recognition : D6LH4CNEB800

Fig 15.2 : Recognition: MH01AX7946

15.2 Vehicle Quantization and Equalization

15.3 Checking and Verification of the Statistic

16 List of functions used

17 CODE 17.1 Algorithm to extract characters ( extractdigit.m).

17.2 Algorithm of Normalisation of characters (same_dim.m)

17.3 Algorithm to read letters (read_letter.m)

18 Result Analysis and Discussion 18.1 Experimental results

Extracted Number plate: MH12EM2587

Extracted Number plate: MH12FF7461

Extracted Number plate: MH12DM3909

Extracted Number plate:LMH12CD9736

Extracted Number plate:LK00A53N6523

Extracted Number plate: MH12HN0389