Deskew after OCR or import

DEVONthink Tips
1

Hi

I’m a new to DEVONthink and using it now for a few weeks. Primarily I’m using DT to import scanned images from my Fujitsu Snapscan device. So far I’m very happy with the solution. But what I’ve noticed is that DT is deskewing some images after scanning. It looks that not snapscan is doing the deskew but DT. The Image from the scanner, stored as PDF, looks very ok. There is no need to deskew the image. But if I look at the PDF after the OCR process, the image is slightly skewed.

Can I do something that DT is no more deskewing some images. I noticed this behaviour on maybe 5 images out of 100 scans.

Thanks for your advise.
kilico

2

This happens automatically and there is no user interface to switch it off.

3

If there is no user interface to configure the behavior of the deskewing, is there any other way to do that?
I noticed that the OCR process isn’t working well on such images. The OCR process produces much less words then on a comparable image without the faulty deskewing.
Maybe there is any other way that i can get such images properly into DT.

4

The OCR built in DEVONthink Pro Office is based on the fire-and-forget principle. If you want full control over the process we advise you to get the apps sold by Abbyy or IRIS that are specialised for this function.

5

Does someone has experience with the full installation of ABBYY Finereader or IRIS in conjugation with DT. What annard was advising is that I should use the full version of ABBYY or IRIS to get better results out of the OCR process.

What I dislike with the results that I get with the included OCR engine, is that the resulting image in the PDF is not faithful to the original document.

A simple solution would be that I can switch off the deskewing in the OCR process. With my Fujitsu 1500 scanner the documents are scanned correctly and they don’t need do be deskewed.

6

Readiris was bundled with my ScanSnap bought 18 months ago; might not be the case today. For deskewing I configure ScanSnap Manager to scan to Readiris, do the deskew and OCR there, and import to DTPO. I use this on scans of very old newspaper articles. To my eyes, the OCR in DTPO2 is better than Readiris, even when skewed. Don’t know about ABBYY FineReader.

7

I like ABBYY Finereader much better than ReadIris.

I wish I could use it outside of the DEVONthink environment with my ScanSnap S300M.

8

I’ve scanned thousands of pages with my first-generation ScanSnap, probably nearly a thousand with ABBYY OCR in DTPO2.

I don’t see skewing of the images in searchable PDFs unless there had been misalignment in page feeding.

Could you send a sample PDF before and after OCR to Support?

9

Many thanks for all your replies.
Because all the documents are bank statements I cannot send you the original pdf’s. But i will make some screenshots and black out some parts for privacy reasons.
Hope you can give me some advice.