DarkenMyPDF

Remove noise from a scanned PDF

Dust on the glass, a dirty drum, a worn original or heavy compression leaves small dark specks scattered across a scan. They are not part of the document, and they print as dirt.

This tool clears isolated specks from the paper and does nothing else. It does not change the tone of the page, the colour of the paper or the ink. A speck is only removed if it is clearly darker than its surroundings and almost entirely alone, which is why strokes are never touched.

Remove noise from your scan

Pick a strength, upload your PDF, and compare the result before you download.

  1. Upload
  2. Enhance
  3. Compare & download
Strength

Removes faint specks too. A good first try.

65
OffStrong

or drop a PDF here — one file at a time, up to 50 MB and 100 pages

Free to try • No signup required • Your file never leaves your device

How to remove noise from a scanned PDF

  1. Pick a strength above. Standard is selected to begin with.
  2. Choose your PDF, or drop it on the page. It is processed in your browser and never uploaded.
  3. Zoom in on the paper between lines of text and drag the slider to compare the original with the result.
  4. If specks remain, choose Strong. If fine detail has gone, step back to Light, then press Apply new settings.
  5. Download the new PDF. The original is left untouched.

What counts as noise here

In this tool, noise means isolated specks: small dark dots on the paper. A pixel is treated as dirt only if it is clearly darker than the median of its eight neighbours and has at most one dark neighbour of its own. A speck of dust has none. A stroke of text, even a thin one, has at least two along its direction, so it is left alone.

The strength slider lowers the bar for "clearly darker". A higher setting catches fainter specks; it does not half-erase more pixels.

What it does not remove

  • Grain across the whole page, such as high-ISO camera noise. There are no isolated specks to find.
  • JPEG blockiness and ringing. These are patterns of the compression, not specks.
  • Halftone patterns and moiré from scanning printed images.
  • Show-through from the other side of a thin sheet. Whitening the paper helps there; see whiten PDF background.
  • Specks that touch the text, and specks larger than a pixel or two at 200 DPI. They look like part of a stroke, so they stay.
  • A one-pixel hairline loses a pixel at each end, because an end has only one dark neighbour. Strokes of normal text are unaffected.

Noise removal and the rest of the cleanup

Which pages are changed

Pages that contain images, including every scan, are rendered at 200 DPI, cleaned and written back as images. Pages made only of text and vector graphics contain no scanned specks, so they are left untouched, and the result panel tells you when that happens.

Questions

Will removing noise blur my text?

No. The method does not smooth or average pixels. It only replaces isolated specks with the surrounding paper, so strokes keep their edges.

Why are some specks still there?

They are probably faint, larger than a pixel or two, or touching a letter. Try Strong. Specks that look like part of a stroke cannot be told apart from the stroke and are left in.

Can it remove grain or compression artifacts?

No. Those are patterns spread across the page, not isolated specks, and removing them would mean smoothing the text as well.

Does it change the colour or brightness of my scan?

No. Only the specks are replaced. Everything else on the page is left as it was.

Will the text stay selectable?

On pages made only of text and vector graphics, yes, because they are not changed. A page that contains any picture, including a scan, is processed as an image, and its text becomes part of that image.

Is my PDF uploaded?

No. Reading, processing and writing the new file all happen inside your browser tab.

Guides on this topic

Related tools

Remove noise from your scan

Free, no signup, no watermark. Your file stays on your device.

How it works · Features · Privacy

Upload scan to remove noise