OCR Advanced (FineReader): Disable Dictionary-Based Correction

[MikeTomsen]

Hello,

I’m using PDF-XChange Editor’s (10.7.3) Advanced OCR (FineReader) for German texts with many proper names and encounter the following issue:

Problem
Advanced OCR applies dictionary-based correction that changes clearly recognized characters. For example, “l” in proper names is often converted to “i.”

Workaround
When I set the recognition language to Latin or Spanish, letters are recognized correctly—but German umlauts (ä, ü, ö) and “ß” are lost.

Standard OCR recognizes all characters correctly on high-quality images but doesn’t offer “Fine Page Content.”

Questions

Can dictionary-based correction be disabled in Advanced mode?

Is there an alternative Workaround?

Can “Fine Page Content” preserve the original page as a layer that can be toggled on later?

Best regards
Mike

[Daniel - PDF-XChange]

Hello, MikeTomsen

There is no way to disable dictionary correction, however there may be a workaround:
Does the same issue happen if you enable *both* the German language, and a Latin language? When you do so, the order in which you select them acts as a priority system, but allows both recognition functions to work in tandem.

And finally, no, fine page content cannot currently preserve the original page content in that way. I can check with the Dev team to see if such a feature is even possible, as I think this may be the first time I have seen such a suggestion.

[A quick update, in the meantime, you can use the "Overlay pages" tool on the organize tab, and specify the original document before saving the OCR output), to be added as a new layer, and then hide that layer manually. This should accomplish what you need with a few quick extra steps].

Kind regards,

[MikeTomsen]

Hello Daniel,

Thank you for the feedback and your advice.

Unfortunately, changing the order of the languages ​​doesn't solve the problem.

Only when I deselect German and select any other language with Latin letters, no dictionary correction takes place, and the characters are recognized correctly – except for the now missing German umlauts ä, ü, ö, and ß.

Apparently, the original Finereader software offers the option to create a custom dictionary and deactivate "Dictionary" in the options so that only the language's alphabet is used.

But my problem is very specific, as I have a text with mostly proper nouns. Therefore, I'm using your suggestion with the layers:

Layer 1: Original pages (I run standard OCR on this, and all letters and words are recognized correctly)

Layer 2: "Enhanced - Fine Page Content" OCR. This significantly improves readability, and I can live with the low error rate due to the dictionary corrections.

The main thing is that I can find the correct entries using the level 1 text search.

Best regards,
Mike

[Daniel - PDF-XChange]

Hello, MikeTomsen

I am glad to hear you have a "working" solution, even if it is not ideal.

Would you perhaps be able to share the original files, and a screenshot of the problematic OCR settings configuration with us here, so we can investigate on this end and see where we can make improvements to the process?

Kind regards,