Difficult Images

[Gordon Cheung]

Hi, I'm currently trying to get OCR to recognize images that I am taking from a camera, from small pipes. Is there anyway to recognize images like these? I'm currently getting no output in the generated Text file. I can convert edit the images if need be, but in which way will allow Tesseract to recognize the single letter? All of the images that I will be generating will look like this one. Just one image inside of a circle.

Thanks for any help!

[David Cuenca]

Have you tried to do some pre-processing with http://opencv.org/ ?

Cheers,

Micru

On Tue, Aug 19, 2014 at 1:20 AM, Gordon Cheung <itsgor...@gmail.com> wrote:

Hi, I'm currently trying to get OCR to recognize images that I am taking from a camera, from small pipes. Is there anyway to recognize images like these? I'm currently getting no output in the generated Text file. I can convert edit the images if need be, but in which way will allow Tesseract to recognize the single letter? All of the images that I will be generating will look like this one. Just one image inside of a circle.

Thanks for any help!

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-oc...@googlegroups.com.
To post to this group, send email to tesser...@googlegroups.com.
Visit this group at http://groups.google.com/group/tesseract-ocr.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/3eae5643-055a-4267-9a84-cb7c1e44a15f%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
Etiamsi omnes, ego non

[Art W Rhyno]

Hi Gordon,

I ran into something like this once with a set of images from microfilm. It might not work on all of your images but one thing you might try is to use gimp for preprocessing. You can script gimp if there is a set of steps that achieves a desired result and in this case, try opening the image, reducing it by 75%, do a color selection based on black, and copying the results into a new image. I think you would often get a result like the attached. If so, you might be able to eliminate the outside color and leave the letter on its own. If the images have the letter in close to the same position and size then you might be able to extract the letter based on coordinates.

art

--

You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to

tesseract-oc...@googlegroups.com.

To post to this group, send email to

tesser...@googlegroups.com.
Visit this group at http://groups.google.com/group/tesseract-ocr.

To view this discussion on the web visit

https://groups.google.com/d/msgid/tesseract-ocr/3eae5643-055a-4267-9a84-cb7c1e44a15f%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.[attachment "A1.jpg" deleted by Art W Rhyno/artrhyno/University of Windsor]