How to adjust the text angles of scanned documents / Texte geraderichten

[Panther]

How to adjust the text angles of scanned documents?
How do you do that for a bunch of documents in praxi?
Background: The adjustment improves the OCR detection rate.


Wie richtet man Texte eingescannter Dokumente aus?
Wie macht Ihr das bei einem ganzen Stapel eingescannter Dokumente in der Praxis?
Hintergrund: Die Geraderichtung erhöht die OCR-Erkennungsrate.

Zur OCR verwende ich bevorzugt Tesseract, weil es prima im Batch-Modus funktioniert.

Thanks for your time. Best,
Panther

[Bhikkhu Pesala]

Assuming that a multipaged scanned document is likely to have to text misaligned differently on every page, the only way is to the use the Straighten Tool of the IrfanPaint Plugin on each page.

If the angle is consistently wrong, then you can use a batch process to apply a custom fine rotation.

[Panther]

Sorry, you are always putting me in the rain ...

Where is a straiten tool? You do not mean [Ctrl-U] "Custom/Fine rotation"?
And, of course, I meant single page graphics.

Panther

[Bhikkhu Pesala]

Open the IrfanPaint Plugin with F12. The Straighten tool is on its toolbar.

[Panther]

Okay, found.

Thank you. It may be better than measuring and working with "Custom/Fine rotation" ([Ctrl-U]), but I would have to use the buggy paint tool (freezes IrfanView) and I hoped someone could tell me an alternative software. The commercial and pretty expensive Abbyy Finereader has a function built in to adjust a text graphics automatically before the OCR process. Finereader has also another disadvantage.

What software could be used for straightening text graphics automatically, best by batch?

Panther

BTW: This function may be named "deskew"/"de-skew".

[Bhikkhu Pesala]

Save a whole batch of images to a PDF file, then run it through PDF-XChange (the free version will do this)

[Panther]

Sorry, but that is nonsense. What is "run it through PDF-XChange"? It can only turn pages in whole 90 deg. portions, and you have to choose them manually.

[Bhikkhu Pesala]

Sorry, but that is nonsense.

I won't waste any more of my time then. If you follow the link you can figure it out for yourself.