SourceForge has been redesigned. Learn more.
Close

#3 JPG|PNG to monochrome TIFF, but nicely

closed
nobody
None
5
2007-05-17
2006-09-05
No

I'm prepared to live with the fact that tesseract can't
convert JPG or PNG. If I convert a picture taken with
my crappy digital camera of an invoice and convert that
to a monochrome TIFF one can barely read any of the text.

There's got to be a way to prepare the image for a
monochrome conversion without loosing too much of the
quality.

Anyone?

Discussion

  • Derek Pomery

    Derek Pomery - 2006-09-05

    Logged In: YES
    user_id=890236

    From personal experience with OCR software, you won't be
    able to get decent character recognition with even a
    non-crappy digital camera. Buy a scanner. A decent flatbed
    is cheap, easier to use, and gets much better results.

    If you want to play with GIMP, netpbm and imagemagick and
    try and turn a fuzzy greyscale into sharp lines, good luck...

    Perhaps you could try hacking up the algorithms in some
    bitmap to vector program like autotrace or potrace - I'll be
    it'll be a lot harder than just buying a proper scanner.

     
  • Peter Bengtsson

    Peter Bengtsson - 2006-09-05

    Logged In: YES
    user_id=208160

    Reply to kyberneticist,

    It doesn't have to be perfect. I just need to get some out.
    It will allow me to guessestimate if the image is rotated
    right or if it even contains any letters.

     
  • JetsoftDev.com

    JetsoftDev.com - 2006-09-16

    Logged In: YES
    user_id=1599597

    You would be better off scanning the invoice with a
    flatbed or handheld personal scanner.

    A digital camera has all kinds of perspective issues.

    The other key thing in converstion would be to increase
    the resolution of the picture first and then convert it.

     
  • Filip Gieszczykiewicz

    Logged In: YES
    user_id=37894

    You want to install NetPBM on your machine. That's what I
    use and it works great. See http://netpbm.sourceforge.net/

    I can't see how you'd survive using tess without it!

     
  • JetsoftDev.com

    JetsoftDev.com - 2006-10-27

    Logged In: YES
    user_id=1599597

    There are ways but these are things you can do in photoshop not tesseract. A camera has 3d perspective. It is probably not the best choice. A portable
    flatbed scanner is a better bet.

     
  • Nobody/Anonymous

    Logged In: NO

    If you have ImageMagick you may test with this:

    convert image.png -monochrome -enhance image.tiff

     
  • Ray Smith

    Ray Smith - 2007-05-17
    • status: open --> closed
     

Log in to post a comment.