Working with scans
How to edit text in a scanned PDF and check the result
In short
A scan stores printed words as pixels, so first recognise the text before editing it. Check the recognised wording against the image, make the intended correction and inspect the saved page. Adding a text layer alone does not erase the original printed words.
What is actually being edited?
A scanned page is a picture. Its printed sentence is not automatically a paragraph object that an editor can reflow. OCR identifies candidate characters and their positions. Editing then needs to deal with both the replacement wording and the original image behind it.
Open the scan with OCR PDF or the appropriate recognition path in PDF Editor. Keep the original unchanged while you work. If you are not sure it is a scan, use the editability diagnosis first.
How reliable were the recognised words in our test?
We recognised a controlled scan with KovaPDF's image tool. It returned four text lines, including Invoice total 125.50 and Reference 4837. The source PDF had zero extractable text runs before recognition.
We compared the amount and reference with the original image. Those values were correct in this fixture. This does not establish an accuracy percentage for other scans, handwriting or languages. Treat the recognised words as a draft until you have checked the characters that matter.
How should I make a small correction?
- Recognise the relevant page and check the proposed text.
- Select the intended wording rather than a neighbouring line.
- Enter the approved replacement and review its length and position.
- Export a separate working copy.
- Reopen the copy and inspect both its appearance and extracted text.
A longer replacement may not fit the original space. Check adjacent lines and the right margin. A convincing visual patch can still overlap a digit or leave part of the old word visible around its edges.
Does OCR remove the old words?
No. A searchable text layer can sit over the same original picture. You may now select the words while the printed image remains exactly where it was. Similarly, placing white over a word and adding new text is not proof that the underlying content has been securely removed.
If the job is confidential removal, use the dedicated removal workflow and verify the result for that purpose. If the job is an approved correction, preserve the source and a record of what changed. A visual edit and a secure redaction are different requirements.
What if recognition keeps getting a digit wrong?
Inspect the original at higher zoom and compare the surrounding context. Improve the capture if the digit itself is ambiguous. The scan cleanup guide explains gentle contrast and noise adjustments, while scan preparation identifies defects that require recapturing.
Do not infer an account number or amount from what seems plausible. For financial or identity data, ask for a clearer source when the image cannot settle the character.
When is a source document better?
Use the editable source when you need substantial paragraph changes, new tables or consistent typography across many pages. Rebuilding a whole report from a scan can introduce more layout and recognition errors than correcting the source and exporting again.
If all you need is the wording outside the PDF, copy recovery is simpler than editing the page. Keep the distinction clear: extracting text, making it searchable and replacing the visible wording each has a different output to verify.
Common questions
Does OCR alone change the printed wording?
No. It can provide text while the original page image remains. A visible replacement needs its own editing step.
What should I check before accepting an OCR correction?
Compare names, amounts and references with the source, then inspect the replacement position and saved output.
Is covering a word with white a secure removal?
No. A visual cover is not proof that underlying content has been removed. Use a removal workflow when confidentiality is the goal.
When should I request the original editable document?
For substantial wording or layout changes, the source is usually easier to correct consistently than a reconstructed scan.
KovaPDF on your computer and phone
Use KovaPDF on your computer, where many tools work with no internet, and on your phone. The same account works in both.