License Plate Recognition System

ABSTRACT:

License plate detection is widely considered as a solved problem with

many systems already in operation. However, the existing algorithms or systems

work well only under some controlled conditions. There are still many challenges

for license plate detection in open environment, such as various observation angles,

background clutter, scale change, multi-plate, uneven illumination, and so on. In

this paper, we propose a novel scheme to automatically locate license plate by

principal visual word discovery and local feature matching. Observing that

characters in different license plates are duplicates of each other, we bring in the

idea of bag-of-words (BoW) model popularly applied in partial-duplicate image

search. Unlike the classic BoW model, for each plate character, we automatically

discover the principal visual words characterized with geometric context. Given a

new image, the license plates are extracted by matching local features with

principal visual words. Besides license plate detection, our approach can also be

extended straightforward to detection of logos and trademarks.

Image Capturing

Extraction of plate regionand

Edge detection

Image Segmentation

Character recognition of Plate

Due to the invariance virtue of SIFT feature, our method can adaptively deal with

various changes of license plate, such as rotation, scaling, illumination, etc.

Promising results of the proposed approach are demonstrated with experimental

study in license plate detection.

Block Diagram:

EXISTING SYSTEM:

License plate detection is widely considered as a solved problem

with many systems already in operation. Nevertheless, the existing algorithms or

systems work well only under some controlled conditions. For instance, some

systems require sophisticated video capture hardware possibly combined with

infrared strobe lights, or require the images taken with little distortion from view

point changes. Although many reported results are very good, with even perfect

accuracy on their test datasets, it is still a challenge task to detect license plates in

open environment.

DISADVANTAGES OF EXISTING SYSTEM:

Generally, a license plate detection system has to solve two problems:

where a license plate is located and how big it is. Usually, the candidate position of

characters in license plate is first identified, and the bounding box of the license

plate is determined later. There are many challenges in license plate detection in

open environment, such as various observation angles from cameras, background

clutter, different sizes of license plate, poor image quality from the uneven lighting

condition, and multi-plate detection.

PROPOSED SYSTEM:

In this paper, we experiment on two types of license plate. It is straightforward to

make extension to detection of other types of license plates in other regions or

countries of the world. Our main contributions include:

1) We propose an automatic method to discover the principal visual words for each

character in license plate. Each principal visual word is characterized with both

local descriptor and some other geometric clues.

2) Based on visual matching with principal visual words, we propose an efficient

scheme to accurately locate the image patch containing license plate.

ADVANTAGES OF PROPOSED SYSTEM:

In this paper, we propose an automatic approach for license plate detection in open

environment. Our approach is based on principal visual word discovering and

visual word matching. We identify the principal visual words of each character.

With the invariance merit of SIFT feature, our approach is effective in dealing with

various observation angles, scale change, and illumination variation, etc. Besides,

our approach can detect multiple plates in an image automatically. Promising

results are shown on two evaluation datasets.

MODULES: Principal Visual Word Generation

Visual Word Matching

License Plate Locating

PLATEREGION EXTRACTION:

SEGMENTATION:

MODULES DESCRIPTION:

Principal Visual Word Generation

In license plates, there are a certain number of sorted characters, each with the

same format but maybe undergoing illumination change or affine transformation.

Since SIFT feature is invariant to changes in scale and rotation, and robust to

illumination change and affine distortion [2], there exist some repeatable and

distinctive SIFT features to each character, called principal visual word (PVW).

Visual Word Matching

Given a test image, we will discover those characters with features matched to the

principal visual words. We first extract SIFT features from the test image. Then

each SIFT feature is compared with the principal visual words of each character. A

feature is considered as a candidate match if the minimum descriptor distance to a

certain principal visual word of a certain character is less than a constant threshold.

License Plate Locating

Once the character features in the test image are identified, we can make use of the

geometric context of the matched principal visual words to locate the license plate.

A bounding box will be estimated to encompass license plate by determining the

upper, lower, left and right bounding lines sequentially.

PLATEREGION EXTRACTION:

Plate region extraction is the first stage in this algorithm. Image captured from the

camera is first converted to the binary image consisting of only 1’s and 0’s (only

black and white) by thresholding the pixel values of 0 (black) for all pixels in the

input image with luminance less than threshold value and 1 (white) for all other

pixels. Captured image (original image) and binarized image are shown in Figures

Fig.1 Captured Image

Fig. 3 Locations of plate characters

Fig. 3 Number Plate Image

SEGMENTATION:

In the segmentation of plate characters, license plate is segmented into its

constituent parts obtaining the characters individually. Firstly, image is filtered for

enhancing the image and removing the noises and unwanted spots. Then dilation

operation is applied to the image for separating the characters from each other if

the characters are close to each other. After this operation, horizontal and vertical

smearing are applied for finding the character regions.

HARDWARE REQUIREMENTS

Processor : Any Processor above 500 MHz.

Ram : 128Mb.

Hard Disk : 10 GB.

Compact Disk : 650 Mb.

Input device : Standard Keyboard and Mouse.

Output device : VGA and High Resolution Monitor

SOFTWARE REQUIREMENTS

Operating System : Windows XP.

Coding Language : JAVA