Tesseract newbie - No output from tesseract

[Tenzin]

Hi,

I am new to tesseract. To test and play around with tesseract, I
created an english document with a few sentences in my wordprocessor
(openoffice.org writer) and then using the printscreen command in
gnome, i created a screenshot of the document. I edited this
screenshot using gimp to just select the text part of the image and
saved it as uncompressed TIFF format (.tif). Using this file as my
input, i ran tesseract using the following command:

tesseract test.tif test -l eng

The program terminates by outputting the text "Tesseract Open Source
OCR Engine". It doesnt show any error. The output text file test.txt
is also created but it is just 1 byte in size and doesnt have any
content in it. I tried creating the test image file using different
types of fonts but the same thing happens. I would therefore like to
know if I am doing something wrong here.

Please help. Also, is there an option to make the tesseract
application more verbose or to make it log its details in any file?? I
am using tesseract 2.03 on Debian Lenny.

Regards
Tenzin

[Sascha Schimke]

Hello,

I would think, that your screenshot just has a resolution, which is to
small. I use tesseract with scanned images of 300dpi. A screen
typically only has less than 100dpi. Try to export a pdf file from
open office, then convert this (using photoshop or gimp) to tif. By
this, you can select the amount of dpi, when opening the pdf file gimp
or photoshop.

Regards,
Sascha

[74yrs old]

Yes. Resolutuion should be 300dpi
cheers

[lab]

In my experience, TIFF files sometimes have an alpha layer. The
easiest way to ensure a usable image for tesseract is to do these two
steps (on Debian)

convert test.tif new_test.pbm
convert new_test.pbm new_test.tif

Then try to OCR the file new_test.tif. The convert program is part of
the imagemagick package.

[Ray Smith]

See http://code.google.com/p/tesseract-ocr/issues/detail?id=160

Ray.

[ABB]

Link not found :-(

On Dec 21, 12:02 am, "Ray Smith" <theraysm...@gmail.com> wrote:
> Seehttp://code.google.com/p/tesseract-ocr/issues/detail?id=160Ray.

[Ray Smith]

http://code.google.com/p/tesseract-ocr/issues/detail?id=160

[Tenzin Dendup]

Thanks for all the help. the OCR worked for me. Changing the dpi to 300 helped in getting some output but the recognition was very low. Using the convert program to convert the gimp-created tiff images to pbm and then again using convert to change the pbm files back to tiff and then running tesseract made it work very well (Recognition was almost 100%) .

Once again, thanks all for the help.

Regards

[74yrs old]

Ray,
Seehttp://code.google.com/p/tesseract-ocr/issues/detail?id=160Ray. not found with error message  as follows:
"

Not Found

The requested URL /p/tesseract-ocr/issues/detail?id=160Ray was not found on this server.

On Wed, Dec 24, 2008 at 6:17 AM, Ray Smith <thera...@gmail.com> wrote:

http://code.google.com/p/tesseract-ocr/issues/detail?id=160

With regards,
-sriranga