How to make a scanned document more readable

1. Resolution: can the detail be resolved at all?

At 150 DPI, 8pt text is about 17 pixels tall — enough for a human to read on screen, marginal for a printer and poor for OCR. At 300 DPI it is 33 pixels, which is comfortable for all three.

Resolution cannot be added after the fact. Upscaling a 150 DPI scan to 300 DPI invents pixels; it does not recover the strokes that were never captured. If small print matters and you still have the paper, rescan.

2. Tone: is the ink actually darker than the paper?

This is the part most scans fail, and the part software fixes best. Normalise the lighting across the page, then map the measured ink towards black and the measured paper towards white. Nearly every "unreadable" scan becomes readable at this step alone.

3. Sharpness: has compression softened the edges?

Scanner optics soften edges slightly and JPEG compression softens them further, particularly at the aggressive quality settings that document scanners default to.

A mild unsharp mask restores the definition. The key word is mild: heavy sharpening produces bright halos around every stroke, which looks impressive at 100% zoom and terrible in print. Sharpening also has to come after noise removal, or you simply make the noise crisper.

4. Colour: is it helping or just adding noise?

Colour is worth keeping when it carries meaning — a blue signature, a red stamp, a highlighted row. It is worth discarding when it is only the yellowish cast of old paper, because then it adds file size and visual noise without adding information.

The useful middle ground is to keep colour where the pixel is genuinely saturated and go neutral everywhere else. That preserves the stamp and cleans the paper in one pass.

5. Searchability: is there a text layer?

A scan without OCR cannot be searched, quoted or read aloud. Running OCR after enhancement rather than before it measurably improves accuracy, because recognition engines are trained on clean, high-contrast text.

So the sensible order is: scan well, enhance, then OCR.

6. Accessibility: can a screen reader use it?

A scanned page is an image, and an image without a text layer is invisible to a screen reader. If the document is going to anyone who may use assistive technology, OCR is not optional — it is the difference between a document and a picture of one.

Good contrast helps human readers with low vision too, and it is one of the few improvements that costs nothing to make.