Improve skew correction/Manually correct skew

  • Last Post 4 days ago
evoflo posted this 2 weeks ago

I have a lot of documents with tables which are slightly skewed. I set the correctSkew property to true (and mode to CSM_CorrectSkewByHorizontalLines) but skew is not corrected.

The larger the table is, the worse the Analysis of the tables is. That is Separators are added in cells where they should be "absent" or the top row is not included in the table.

top-left corner of the table      .....           top-right corner of the table

You can see that the separator is not aligned with the border of the cell (because the table is skewed). Since the table takes up 3/4 of the page and I set the mode to CSM_CorrectSkewByHorizontalLines, I expect the skew to be corrected.

I have an external tool which determines the skew angle correctly. Unfortunately, the tool does not take pdfs. Therefore exporting and importing the Image decreases the Quality of the document which in return decreases recognition Quality of the engine.

If I could manually tell the engine which skew angle there is in the document, it would fix the Problem. It would be easier if the engine would find the correct angle itself.

Any other suggestions?


Order By: Standard | Newest | Votes
evoflo posted this 5 days ago

I also noticed that the correctSkew property of the PagePreprocessingParams causes the following issue.

If I load multiple documents, the engine puts a line through text, even though it should be a merged cell:


This does not happen if I do not set the property in the Profile or if I load a single page.

The SplitOnlyBySeparators = true does not help.


Anna Fedyushkina posted this 4 days ago


Please send us some images on which your issue is reproduced. You can either post them here or send them to