visualization of the page structure

Janusz
2013-04-02
2013-04-05
  • Janusz
    Janusz
    2013-04-02

    Jakub Wilk's ocrodjvu stores in the output DjVu files all the information about the page segmentation provided by the selected OCR engine, in particular by tesseract. Unfortunately there is practically no way to have a look at it, with the exception of the lowest level units which are displayed together with the hidden text.

    I would be very happy to have a way to see the whole page structure, as tesseract native tools are very cumbersome (at least those I know). I think it can be done in djview4 at least in two ways:

    1. to add extended/additional hiddent text mode,
    2. to convert temporarily the segment into annotations displayed in the standard way.

    My primary application would be checking tesseract column recognition in a historical dictionary.

    What do you think about it?

    Best regards

    Janusz

     
    • Leon Bottou
      Leon Bottou
      2013-04-05

      Hello Janusz,

      You could experiment by changing function QDjVuPrivate::paintHiddenText()
      in qdjvuwidget.cpp.

      (see
      http://sourceforge.net/p/djvu/djview-git/ci/ba337199479b512c312331274ec6e664
      bf39eb4e/tree/src/qdjvuwidget.cpp) and draw rectangles around all the
      regions instead of simply the terminal one.

      The only problem is that variable Page::hiddenText contains a flattened form
      of the hidden text (see function flatten_hiddentext) for performance
      reasons. You probably want to save an unmodified copy of the hidden text
      in an additional field of the Page structure.

      • L.

      From: Janusz [mailto:jsbien@users.sf.net]
      Sent: Tuesday, April 02, 2013 12:45 AM
      To: [djvu:discussion]
      Subject: [djvu:discussion] visualization of the page structure

      Jakub Wilk's ocrodjvu stores in the output DjVu files all the information
      about the page segmentation provided by the selected OCR engine, in
      particular by tesseract. Unfortunately there is practically no way to have a
      look at it, with the exception of the lowest level units which are displayed
      together with the hidden text.

      I would be very happy to have a way to see the whole page structure, as
      tesseract native tools are very cumbersome (at least those I know). I think
      it can be done in djview4 at least in two ways:

      1. to add extended/additional hiddent text mode,
      2. to convert temporarily the segment into annotations displayed in the
        standard way.

      My primary application would be checking tesseract column recognition in a
      historical dictionary.

      What do you think about it?

      Best regards

      Janusz

      visualization of the page structure
      https://sourceforge.net/p/djvu/discussion/103286/thread/77392e32/?limit=25# ca7d


      Sent from sourceforge.net because you indicated interest in
      https://sourceforge.net/p/djvu/discussion/103286/

      To unsubscribe from further messages, please visit
      https://sourceforge.net/auth/subscriptions/