Log in to create and rate content, and to follow, bookmark, and share content with other members. Not a member? Join Now!

AnsweredAssumed Answered

OCR pdf have some shift of selection bug

Question asked by djarty on Jan 26, 2018
Latest reply on Feb 7, 2018 by djarty

Like • Show 0 Likes0
Comment • 10

Hello!

Looks like Alfresco (Community 201707) have some trouble with searchable word selection into OCRed pdf files?

See picture.

See some shift of selection, not really under the real words and letters. Its unusable if need select and copy/paste some word from OCR pdf.

Its a bug (please confirm somebody) or can be tuned for normal look?

On picture also Chrome and Adobe Reader window with same file and normal selection directly under letters. So its looks like not OCR engine problem but Alfresco rendering.

Attachments

TestOCR.pdf26.0 KB
Selection.jpg162.6 KB

No one else has this question

Outcomes

10 Replies

krutik.jayswal Jan 26, 2018 12:55 PM
Mark Correct
Correct Answer
Its the bug of the tool which you are using and not of alfresco.For example , if you are using pdfsandwich for performing ocr than its bug of pdfsandwich. Not sure which tool you are using.

Basically when you perform the OCR, what every tool does is creating a text from the pdf/image and put it on behind the what we visually looking at.So it often happens that the coordinates become bit of wrong.

OCR depends on the quality of image as well.It is totally depends on the capability of the OCR Tool.
Like • Show 0 Likes0
Actions
- djarty @ krutik.jayswal on Jan 26, 2018 2:06 PM
  Mark Correct
  Correct Answer
  My first thinking - ocr tool problem (yes its sandwich). But I specially make two other windows with Chrome pdf rendering engine and Adobe Reader engine with same pdf. As you can see shift problem only in Alfresco Document Details view.
  
  I will try just add clear pdf maked in other ocr (if find it) or directly from pdf maker. But think problem not gone.
  Like • Show 0 Likes0
  Actions
  - krutik.jayswal @ djarty on Jan 26, 2018 2:13 PM
    Mark Correct
    Correct Answer
    Can you upload same pdf document in forum..So that i can check it in my system....Thank you
    Like • Show 1 Like1
    Actions
    - djarty @ krutik.jayswal on Jan 26, 2018 2:19 PM
      Mark Correct
      Correct Answer
      Done. Thanks
      Like • Show 0 Likes0
      Actions
mehe Jan 26, 2018 1:35 PM
Mark Correct
Correct Answer
As you showed with your screenshots, it seems to be a problem with the pdf viewing component in Alfresco. The Browser PDF Components seem to be aligned better.
What OS and Version of Alfresco are you running? Maybe it's a problem with the installed Fonts...
Like • Show 1 Like1
Actions
- djarty @ mehe on Jan 26, 2018 2:21 PM
  Mark Correct
  Correct Answer
  Community Edition 201707 GA Ubuntu 16.04 (server and client)
  Yes, browser show all right.
  Adobe Reader - ok.
  
  Usability of selection poor - select visually one word but really got word from upper string if strings have little space..
  OCRed pdf added, please try it on other system, viewer or Alfresco.
  
  P.S. forgot.. the same bad view in Opera (also Chrome engine) in Afresco under Windows OS client.
  Like • Show 0 Likes0
  Actions
djarty Jan 29, 2018 6:00 AM
Mark Correct
Correct Answer
So anybody can confirm selection problem in OCRed pdf? (TestOCR.pdf in first message)
Like • Show 0 Likes0
Actions
- mehe @ djarty on Jan 29, 2018 6:06 AM
  Mark Correct
  Correct Answer
  yes,
  I also have the problem with the "offset" when select Text of your PDF
  1 person found this helpful
  Like • Show 0 Likes0
  Actions
  - djarty @ mehe on Jan 29, 2018 9:33 AM
    Mark Correct
    Correct Answer
    Ok, thank you, its about "my pdf". What about another "correct pdf" may be you have it?
    Just need to prove its global or local problem.
    Like • Show 0 Likes0
    Actions
djarty Feb 8, 2018 7:48 AM
Mark Correct
Correct Answer
Hello!
Ok, make some changes.. Now possible to use tesseract 4.00.00alpha under ocrmypdf or pdfsandwich
And selection in Alfresco now is little bit normal.

But, selection lost the spaces between the words (in Alfresco, in other viewers ok).
What can do with this thing?
Like • Show 0 Likes0
Actions

OCR pdf have some shift of selection bug

Attachments

Outcomes

Related Content

Recommended Content