car plate recognition

Question

3.00/5 (2 votes)

See more:

any idea how to extract car plate number from jpeg image??

i trying to write this code but i need some assistance.....
mind to share with me...a source code would be nice....[sigh]

Posted 15-Feb-06 1:14am

jianloong

Add a Solution

3 solutions

Solution 1

Well, most car plates are done in a standard font

If you could create an array of images / vectors showing the framework for each possible character

(ie the skeleton of an F would be just the basic lines), and then had it search for this shape within each letter in the license plate.

What you have to allow for are lighting, the angle at which the photo was taken, possible different sizes, speeds of car etc.

Looking for a framework will allow you a pretty accurate measurement, and you can make countermeasures against specific problems (ie images on license plates etc)

Regards
James

Posted 15-Feb-06 1:55am

James Gupta

Comments

#realJSOP 8-May-10 11:06am

Assuming all plates use a standard font is not valid. Texas currently has three different license plates, all of which use a different font.

Solution 4

Hi jianloong
follow this
1- find out how many color the plate has ,assuming it is cosist of 2 red and white,then redraw your image on black backcolor
2- most likley the plate will not be balanced i mean it should be tilted or rotated at some small angle so you have to rotatet it back untill you get the width aligne with horizontal line
you may need to skew it too
3- you should know in advance the dimension of the plate ,using this data you can scale the plate to its normal dimension you can realize how far is the car from your camera unless if you have aradar
4- assuming the plate has 3 digits and 3 character,u should divide the width to 6 (3digit +3 char ) segment i will refere to this as 6 sprits
5- for each sprit store the data in 2D more likley char array
6- remember the inner color of the plate is red so we should be searching for each pixel that has a color red
7 knowing the color of the pixel we can easily calculate its location x,y
8 you should have data stored for all digits and data which store all alphabtic
9 - you need to compare each sprite with those stored data
10- you should be carfull that light is not alawys at same concentration
so this will accumulate errors to your data ,forget about night work
11- make sure the car will not ESCAPE before your slow calculation
so you should be selective in chosing the wright language and the best street i do not recomend highway this camera should be in low speed street
12- the most important thing is the likleyhood of wrong result and obtainning wrong number of other cars . which i beleive it should be tangent to your program
you need alot of research
khalid sabtan
saudia arabia

11-

Posted 8-May-10 4:48am

khalidSabtan

Add a Solution

Add your solution here

Treat my content as plain text, not as HTML

Preview 0

…

Existing Members

Sign in to your account

...or Join us

Download, Vote, Comment, Publish.

Your Email
Password
Forgot your password?

Your Email
This email is in use. Do you need your password?
Optional Password

I have read and agree to the Terms of Service and Privacy Policy
Please subscribe me to the CodeProject newsletters

When answering a question please:

Read the question carefully.
Understand that English isn't everyone's first language so be lenient of bad spelling and grammar.
If a question is poorly phrased then either ask for clarification, ignore it, or edit the question and fix the problem. Insults are not welcome.
Don't tell someone to read the manual. Chances are they have and don't get it. Provide an answer or move on to the next question.

Let's work to help developers, not make them feel stupid.

Pete O'Hanlon · Accepted Answer · 2010-01-12T02:37:00

Unfortunately, you aren't going to be given the code to do this as this is a hugely complicated area, with commercial systems going for huge amounts of money. Typically ANPR systems make use of several stages to identify the number plate, these stages being:

1. Plate location. This step is responsible for identifying where the plate is in a picture.
2. Plate translation. Orienting and sizing the image, compensating for skew.
3. Image normalisation. Adjusting the brightness/contrast of the image.
4. Segmentation. Finding the characters themselves (i.e. removing items such as flags).
5. Optical Character Recognition.

Now, ANPR requires cameras to take pictures at very high speeds, and requires the processing to take place very quickly, which means that .NET is probably not going to be the most appropriate choice for processing the image (which typically requires sub-second processing). Typically these systems are written in C++ or C, are highly optimised and make use of image processing techniques such as edge detection and sharpening filters.