OCR works from page image geometry as well as characters
OCR does not only identify individual letter shapes. It also has to segment the page into lines, words and regions so the recognised characters can be assembled in the correct order.
When text lines are noticeably tilted, that layout analysis becomes harder. Tesseract documentation specifically warns that excessive skew can significantly reduce line-segmentation quality.
Straighten the scan before recognising text
If the pages share a small consistent tilt, straighten the PDF first and then run OCR on the corrected copy. The aim is to make text lines horizontal enough for the OCR engine to analyse them cleanly.
ABBYY also treats skew correction as document-image preprocessing and describes deskewing skewed scanner or camera images to improve OCR quality.
Do not confuse deskew with 90-degree rotation
Deskew addresses small angular errors such as a page leaning 2 degrees. A page that is sideways or upside down needs orientation correction instead.
Use Rotate PDF for quarter-turn orientation problems, and Deskew PDF for fine tilt correction.
One angle is best for consistently skewed documents
1into1 estimates tilt from the first-page preview and applies the selected angle across the PDF. That is appropriate when a scanner feeder produced a consistent lean across the batch.
If every page has a different tilt, verify the output carefully because a single correction cannot independently deskew each page.
Run OCR only after the page alignment looks right
After straightening, open the corrected PDF and inspect several text lines. Then use OCR PDF to create the searchable text layer.
Deskewing does not guarantee perfect OCR. Resolution, contrast, noise, font shape and scan quality still affect recognition, so verify important names and numbers against the visible document.
Official sources reviewed
- Tesseract OCR — Improving the quality of the outputDocuments rotation/deskewing as image preprocessing and explains the effect of skew on line segmentation and OCR quality.
- ABBYY FineReader PDF User GuideLists deskewing and page-orientation correction among preprocessing steps used when scan defects reduce OCR quality.