circle.ch :: How to OCR multipage PDF files

circle.ch weblog by Urs Gehrig

A weblog about libre software, law, technology, politics and the like.

2013-06-17T21:58:57

Academic, Android, Apache, Apple, Art, Best Of, Biotech, Civil Society, Content Management, Cooking, Copyright, Creative Commons, Crosspost, Culture, Database, Deutsch, DRM, Economy, Education, Event, Gadget, General, Geodata, Government, Health, Howto, Humor, Innovation, Intellectual Property, Java, Language, LaTeX, Law, Linux, Media, Moblog, Mozilla, Music, Office, Open Content, Open Source, P2P, PHP, Podcast, Politics, Privacy, Projects, Random Thought, Rant, Science, Search, Social Network, Software, Sport, Talks, Technology, Technology Transfer, Travelling, Weblog, Wiki, Wireless and Mobile, XML

22. April 2008

How to OCR multipage PDF files
Open Source (wiki) | Content Management (wiki) - urs @ 16:46:53

The OCR applied here only serves for reasons of indexing PDF files. The page layout will get lost. Nevertheless, the following three steps help you to convert multipage PDF files to a single text file:


 $ convert -density 150 foo.pdf ./tesseract/tmp/p%02d.tif
 $ montage.exe ./tesseract/tmp/*.tif -tile 1x -mode concatenate ./tesseract/tmp/foo.tif
 $ tesseract.exe ./tesseract/tmp/foo.tif output -l eng

For reasons of simplicity the TIF files p00.tif to pXY.tif will get concatenated together to a single TIF file, that has the width of a single page and the height of XY pages. In such a way at least the order of the text or the text flow respectively will be preserved. But one could also concatenate a mosaic of all the TIF files. The density of 150 (dpi) gives reasonable results with tesseract.

Related post(s):

Adding header or footer information to existing PDF (13.6)
PDF creator for Windows applications (6.70)
Personalizing PDF files (6.67)
Click-fix a PDF with Shimbun (6.37)
PDF creation from PS or TIFF files (5.33)

Related tags(s):

software, wiki, humor, open_content, talks

Comments (4) Permalink del.icio.us

The URL to TrackBack this entry is:
http://circle.ch/blog/b2trackback.php/1845

Comment by matthias @ 2008-04-23 10:01:41:
in case you just want to extract text from a pdf, "pdftotext" might be easier. it's part of the xpdf package (http://www.foolabs.com/xpdf/download.html)
greetz, matthias

Comment by Urs @ 2008-04-23 16:12:51:

ThanQ matthias. That is a good point. In some cases, pdftotext will not extract text but binary information and then the above shown method might help. An interesting implementation for Plone using OCR is given by NA. I have tried it out and it works perfectly.

Comment by LuNeX @ 2008-04-29 00:44:34:
Ich weiss mir nicht anders zu helfen, kann leider kein englisch. Gibt es eine Möglichkeit auf deutsch? Mich interessiert wie der QR Code zu jedem Beitrag gemacht wird
Comment by Urs @ 2008-04-29 10:29:30:
Unter http://www.swetake.com/qr/qr_cgi_e.html findet man entsprechende Scripts in PHP.