Welcome Guest! To enable all features, please Login or Register.

Notification

Icon
Error

Options
View
Last Go to last post Unread Go to first unread post
#1 Posted : Tuesday, April 22, 2008 5:09:38 AM(UTC)
Bob van Schie

Groups: Registered
Posts: 4


I'm new to leadtools. I used the OCR samples to process attached image.

These images come from a scanner and the text is allways rotated differently.

Can I , How do I get the OCR recognized words from these kind of images.

Bob.
Bob van Schie attached the following image(s):
image2-300dpicolor.JPG
 

Try the latest version of LEADTOOLS for free for 60 days by downloading the evaluation: https://www.leadtools.com/downloads

Wanna join the discussion? Login to your LEADTOOLS Support accountor Register a new forum account.

#2 Posted : Tuesday, April 22, 2008 5:13:30 AM(UTC)
Bob van Schie

Groups: Registered
Posts: 4


Another sample.

Bob.
Bob van Schie attached the following image(s):
image1-300dpicolor.JPG
 
#3 Posted : Tuesday, April 22, 2008 9:39:59 AM(UTC)

Walter  
Walter

Groups: Tech Support
Posts: 367

Thanks: 1 times
Was thanked: 4 time(s) in 4 post(s)

Bob,

It would be helpful to know what version of LEADTOOLS you are using so I can work with specifically what you have.

If you are using either version 14.5 or version 15 of the Document Imaging Suite, load the OCR 32-Bit demo from either the API or DOTNET. With the demo program, open your image file you want to OCR. Then using the menu option "Pages", insert the current image. This will load your image into the OCR engine. From there, use the menu item "OCR" & then "Recognize Page". This will cause the OCR engine to run, and obtain the text from your image.

Once the image has been run through the OCR engine, you can preview the text using "OCR" -> "Get Recognized Words..."

This will give you a list of the words, but it may not display all of the characters correctly for your image. I recommend using "OCR" - > "Save Result..." to save the text out to a text file. This will give you a better result for the files you supplied, and it will align the text better.
Walter Bates
Senior Support Engineer
LEAD Technologies, Inc.
LEAD Logo
 
#4 Posted : Tuesday, April 22, 2008 9:22:04 PM(UTC)
Bob van Schie

Groups: Registered
Posts: 4


Walter,

i do exactly the things you say, the moment i do a recognize page, you can see the recognized zones. With the about 30% 'rotated' image the zones are squares but not 'rotated' squares. I can do a manual rotation and try to recognize it. My problem is i have to recognize thousands of images a day, they are optionally all in different rotations.

I do have knowledge about the images, i know the allowed words.

Is there a way to tell the engine to try rotating for a optimal result.

On an average images there are 2 to 3 stickers 'white on black' with a barcode and/or text on it.

Thanks,
Bob.
 
#5 Posted : Thursday, April 24, 2008 6:00:09 AM(UTC)

Walter  
Walter

Groups: Tech Support
Posts: 367

Thanks: 1 times
Was thanked: 4 time(s) in 4 post(s)

It would be helpful if you could explain what your vision is in using our OCR engine with your images. You've stated that you have a large number of images that you want processed. Do you intend to manage each image one at a time or were you intending on having an automated process to analyze your images? If so, will you be using the API, C Class Lib, or .NET to program in? What do you intend on doing with the text and/or barcodes in the images?

Using the demos that we provide, you can use the auto-orient feature under Page Process to rotate the specified page to be readable and update the image data. The auto-orient feature can be used in conjunction with manually rotating the image.

If you're not looking to manually recognized each image individually, it may be possible to programmatically locate to barcodes in your image. Once located, rotate the image based on the rectangle returned.
Walter Bates
Senior Support Engineer
LEAD Technologies, Inc.
LEAD Logo
 
#6 Posted : Friday, April 25, 2008 12:01:16 AM(UTC)
Bob van Schie

Groups: Registered
Posts: 4


Walter,

I want to automate the process using .Net , images come from an industrial scanner; purpose is to recognise Brands / Types - and make a descision, what to do with the item (for instance also record a serial number).

The scanner recognises barcode, but if this one is missing or damaged we need to OCR. It is possible that for 1 item multiple 'black/white' images are needed (different sites) before a complete/valid description of the item is put toghether.

There are about 1000 different items, and have to recognise about 15k per day.
 
#7 Posted : Friday, April 25, 2008 11:19:59 AM(UTC)

Walter  
Walter

Groups: Tech Support
Posts: 367

Thanks: 1 times
Was thanked: 4 time(s) in 4 post(s)

The only solution I can think of boils down to trial and error. After the image has been scanned in, it will need to go through a process by which it is rotated a small amount until the proper Brand/Type has been recognized or the image has rotated 360 degrees. Images that fail to be recognized can be cataloged for alternative processing later.

Unfortunately, these images prove very difficult to process for a blind computer. In order to OCR an image, the text needs to be horizontal, and there's not enough information on most of these images to automatically detect and rotate based on the direction of the text or other image features.
Walter Bates
Senior Support Engineer
LEAD Technologies, Inc.
LEAD Logo
 
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.

Powered by YAF.NET | YAF.NET © 2003-2020, Yet Another Forum.NET
This page was generated in 0.079 seconds.