License Plate Recognition

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 22

LICENSE PLATE RECOGNITION

Done By:
Gantyala Ankitha (1085-20-539-023)
Konduka Sanjali (1085-20-539-038)
Musham Aishwarya (1085-20-539-046)
Pandi Tarakalakshmi (1085-20-539-050)
Pydi Sanghavi (1085-20-539-056)
ABSTRACT
License Plate Recognition:
A License Plate Recognition is a computer system which recognizes any
digital image automatically on the license number plate. This method
includes various operations like taking pictures, localizing the number pad,
truncating characters and O C R from alphanumeric characters.
The main idea of this technique is to design and develop effective image
processing techniques and algorithms to localize the license plate in the
captured image, and to divide the characters from that number plate and to
spot each character of the segment by using the Open Computer Vision
Library.
This has been implemented in the K-N N algorithm and python
programming language. Many applications are often implemented by using
this system, like security, highway speed detection, violation of light,
identification of handwritten text, discovery of stolen cars, automatic fee
collection systems.
PROBLEM STATEMENT
Computer vision in Intel’s research has been producing a fruit called Open
Computer Vision (Open CV), which can support the development of
computer vision.
All of those vehicles have a unique vehicle identification number as their main
identifier. The I D is actually within the license number that refers to a legal
license to participate in the public movement.
The auto license plate recognizing system replaces the manual license plate
number writing process within the computer system.
It involves three major steps. They specify number pad character segmentation,
and character recognition.
One of these methods used the adaptive threshold to highlight the characters
and suppress the background. So as to get rid of unwanted image spaces, a
component algorithm is first applied to the converted binary image from the
original panel. Some A N P R system applications are automatic traffic control
and tracking system, highway toll collection / automatic parking systems,
petrol station automation, flight time monitoring. These systems automate
the method of identifying vehicle license numbers, making it fast, cost
effective.
OBJECTIVE
The main objectives of this project are –

• To regulate the traffic automatically


• It helps in tracking system
• To supply fast highway toll collection
• It’s considerably useful in Automatic parking systems
• Petrol station automation
EXISTING SYSTEM
The existing system to determine the details of a vehicle requires a lot
of human interaction which also leads to human errors.
For example, the e-challan fined by the traffic officials is as follows. A
C C T V camera continuously records footage of the ongoing traffic. If a
motorist breaks any traffic rule, the act will be recorded in the footage.
The police will try to extract the number from the vehicle’s screen-shot
captured from the C C T V footage and the offense will beregistered in the
records.
This process requires a lot of human resource which can be reduced
by using the following algorithm to recognize the registered vehicle
number.
PROPOSED SYSTEM
The first step towards Number Plate Detection is database collection which
has been collected from Kaggle.com, one of the most popular and
trustworthy sites which contains a variety of different datasets. This
dataset obtained contains number plates of different cars from different
states having huge variation in their registration numbers.
The next step involves data preparation using the dataset obtained from
Kaggle.
For preparation of our data, we implement the following steps:
• Data cleaning
• Reorder our data according to the pixel density.
• Splitting of data for Training and Testing.
• Create input samples.
• Create target outputs for the training and test sets
• Convert data prepared to .png files.
SYSTEM REQUIREMENTS
Hardware requirements:
The following hardware was used for the implementation of the system:
Processor : Any updated processor
Hard Disk : 100 G B
RAM : 4 GB

Software
requirements :
The following
software was used
for the
implementation of
the system:
SYSTEM
ARCHITECTURE
The following are the steps for the procedure:
1. Firstly, we capture the image and then we use the detectMultiScale
method. This method allows us to detect objects of different sizes in
the input image.
2. And it returns a list of rectangle bounds where objects are
detected.
3. We run a code to detect the car number plate once the car license
plate has been successfully detected and bounded by a red
rectangle.
4. Our next step is to place our emphasis on the license plate itself, and
work towards extracting the number that is there on the license
plate.
DATA
FLOW
DIAGRAM
ACTIVITY

DIAGRA
M
METHODOLOGY
In this section, we tend to study the methods and ways of implementing the framework. We tend to study the
technologies and algorithms used in achieving the objective. In section A we provide information regarding the
used algorithms and technologies. In section B we can justify our projected system.

OpenCV: Open Computer Vision helps in detecting and capturing of data with the help of a camera. OpenCV-
Python is a library of Python bindings designed to solve computer vision problems. It makes use of Numpy,
which is a highly optimized library for numerical operations with a MATLAB-style syntax. All the OpenCV
array structures are converted to and from Numpy arrays. OpenCV (Open Source Computer Vision Library) is
an open source computer vision and machine learning software library. OpenCV was built to provide a common
infrastructure for computer vision applications and to accelerate the use of machine perception in the
commercial products. Being a BSD-licensed product, OpenCV makesit easy for businesses to utilize and
modify the code.
K-NN: In pattern recognition, the k-nearest neighbours algorithm (k-NN) is a non-parametric method used for
classification and regression. In both cases, the input consists of the k closest training examples in the feature space.
The output depends on whether KNN is used for classification or regression.

Its purpose is to use a database in which the data points are separated into several classes to predict the classification of a
new sample point. K-nearest neighbours (k-NN) algorithm uses ‘feature similarity’ to predict the values of new data
points which further means that the new data point will be assigned a value based on how closely it matches the points
in the training set.

Let’s briefly talk about the Open CV detect Multi scale method. The method allows us to detect objects of different sizes
in the input image, and it returns a list of rectangle bounds where objects are detected. For each rectangle, there will be
4 values returned, and they correspond to the following respectively:
• x-coordinate of bottom-left corner of rectangle (x)
• y-coordinate of bottom-left corner of rectangle (y)
• width of rectangle (w)
• height of rectangle (h)
The key parameters involved in the detect Multi Scale function are scale factor and minimum
neighbors. Scale Factor specifies how much the image size is reduced at each image scale (as part of the
scale pyramid, which is a multi-scale representation of an image).

In essence, when object detection models are trained, they are trained to detect objects (i.e. car plate in our
case) of a fixed size, and might miss car plates that are bigger or smaller than expected. As part of the scale
pyramid, the image is resized several times in the hopes that a car plate will end up being a “detectable”
size.

We used the default scale factor of 1.1, which means that OpenCV will scale the image down by 10
percentage to try and match the car plates better. Minimum Neighbors allows us to specify how
many neighbors each candidate rectangle should have, in order for the candidate rectangle to be
retained. In simpler terms, this parameter influences the quality of the detected objects.

A higher value results in fewer detections, but the detections come with higher quality and accuracy. This
means that a higher value can actually help with reducing the number of false positives. For my case, the
default value of 3 gave me several false positives, so I increased it to 5. These two parameters can be
tuned accordingly so as to improve your results.
TEST CASES
TEST CASES
TEST CASES
TEST CASES
CONCLUSION
The result is a method that recognizes the characters from the photo of a car taken from a good
resolution camera and prints the number of the License plate of the car as an output. This project
serves as a stepping stone for larger scale (and more advanced) computer vision projects, such as bulk
extraction of car license plate text from large image quantities, and even application of these
concepts on video hies or live feed. Future extension of this work is to develop character recognition
using template matching algorithms.

Advances technology took Automatic Number Plate Recognition (ANPR) systems from hard to set up,
limited expensive, fixed based applications to simple mobile ones in which “point to shoot” method
can be used. This is possible because of the creation of software which ran on cheaper PC based and
also non specialist hardware in which their no need to give pre-defined direction, angle.

You might also like