Working with scans

How do I edit the text in a scanned PDF?

You cannot edit a scan directly, because the page is a picture. OCR PDF recognises the words and rebuilds them as real editable lines on top of the original page, so you can correct a name or a figure and save a normal PDF.

4 min read

Someone sends you a scanned contract with your name spelled wrong. You open it, click on the name, and nothing happens. There is no cursor, because there is no text — the page is a photograph, and you cannot put a cursor inside a photograph.

The usual advice at this point is to retype the whole document. You do not have to.

What editing a scan actually involves

Three things have to happen, in order, and it is worth knowing what each one costs.

  1. The words are recognised. Optical character recognition looks at the pixels and works out which characters they represent.
  2. They are put back where they were. Not just the text, but the position of every line, so a rebuilt page looks like the page it came from rather than like a wall of recovered sentences.
  3. The original stays underneath. The scan remains as the visual. You are editing a layer of real text sitting on top of it.

That third step is what makes the result usable. The logo, the signature, the stamp, the ruled lines of the form — none of that is text, none of it can be recognised, and all of it stays exactly as it was.

OCR PDF does all three when you open a scanned document, and then behaves like the ordinary PDF Editor: click a line, type, save.

What it will not do

It will not match the typeface perfectly. The original font is not recorded anywhere in a scan — it is ink in a picture. The rebuilt line is set in the closest available face at the measured size and position. On a printed document this is usually unnoticeable in a line of body text and noticeable in a heading.

It will not hide the join on a textured scan. A page photographed on a phone carries shadow, paper grain and slight curvature. Fresh, clean text placed on top of that does not carry any of it. The edit is faithful; it is not invisible.

It will not fix what it could not read. Recognition errors become editable text containing the error. Which is better than being stuck — you can correct them — but it means the page needs reading, not trusting.

Numbers deserve a second look

Recognition mistakes are not random noise; they are confusions between shapes that genuinely look alike. 1 and l, 0 and O, 5 and S, 8 and B, a comma and a full stop.

In a sentence that hardly matters, because you read the word and your eye repairs it. In a total, an account number or a date it matters enormously, and it is exactly where a plausible-looking error does the most damage. A figure that reads 1,500.00 when the paper said 7,500.00 is not a typo anybody notices.

Check every number against the original page before you save. It is on the screen behind your edit, which is the one advantage of this approach.

Get the scan right before recognising it

Accuracy is decided mostly by the input, and most of the input is under your control.

  • Resolution. 300 DPI when you scan. Below about 200 the letters have too few pixels to tell apart.
  • Straightness. A page scanned at an angle is much harder to read. Deskew it.
  • Contrast. Clean black on white. Grey photocopies and coloured backgrounds both hurt.
  • Flatness. A phone photograph of a curved page bends the lines, and bent lines recognise badly. Press the page flat, or scan it.

If what you have is already grey or speckled, Enhance Scan cleans it up first, and recognition afterwards is measurably better.

Which languages

Eight: Bangla, English, Hindi, Spanish, Dutch, Arabic, Chinese and Russian. The language is worked out from the page automatically, and you can name it yourself before uploading if you already know — which is quicker and, on a short page, more reliable than letting it guess.

When retyping really is faster

Be honest about the arithmetic. Rebuilding is worth it when you need to change a little of a lot — a name through a twelve-page agreement, a date on a certificate, a corrected figure on an invoice.

It is not worth it for a handwritten page, for a heavily designed page where the layout matters more than the words, or for half a page of text you could retype in four minutes. General-purpose recognition does poorly on handwriting, and no amount of cleaning changes that.

Common questions

Will anyone be able to tell the document was edited?

On a clean printed scan, usually not at a glance. On a textured or photographed page, a close look can reveal it: the original typeface is unknown and the new text does not carry the paper's grain. It is a faithful reconstruction rather than a restoration.

Does the rest of the page change?

No. Only the lines you edit are redrawn. Logos, signatures, stamps and rules are part of the scanned image and are left exactly as they are.

Can I edit a handwritten document?

Not usefully. General-purpose recognition is built for printed type and does poorly on handwriting, so what comes back needs more correcting than retyping would have taken.

Is the saved file a normal PDF?

Yes. It opens anywhere, and the text you added is real text — selectable and searchable — sitting on the original page.