AuraDoc
Sign inSign up

What is binarization?

Binarization is the step that converts a scanned page into a picture with only two tones, black for ink and white for paper. Each pixel is compared with a threshold: if it is darker than the threshold it becomes black, otherwise it becomes white. The result is a one-bit image, the simplest kind a scanner can produce, and it is the form that text recognition programs often work on internally even when you never see it.

Why it is used and what it costs

A black-and-white page needs one bit per pixel where a grayscale page needs eight and a color page twenty-four, so a binarized scan can be many times lighter. Sharp edges also help a recognition engine tell one letter from another. The cost is that everything between black and white disappears. Pencil, pale stamps, yellowed paper, shadows near the spine of a book and the soft tones of a photograph are pushed to one side or the other. A threshold set too high swallows faint strokes into white, and one set too low turns a grey background into a black block that hides the text.

Choosing the threshold in practice / When to avoid it

Fixed thresholds work on clean, high-contrast printing. Pages with uneven lighting do better with an adaptive threshold, which is calculated for each small region instead of once for the whole sheet. Always look at a sample page at full size before you process a whole batch, and compare thin lines, small digits and stamps with the paper original. Keep binarization for plain text documents such as letters, contracts and forms. Pages that contain photographs, signatures in light ink, hand-written notes in pencil or colored highlights are better kept in grayscale or color, because the lost tones cannot be recovered afterwards.

Using AuraDoc for it

OCR PDF reads the text of a scan and adds a searchable layer, Compress scanned PDF lowers the weight of page images when a scan is too large, and Copy text from scanned PDF gives you the recognized words. Each result is a new file, so your original scan stays as it was.

Tools that use this

Guides that put it into practice

Common questions

Does binarization make the file smaller?
Usually yes, often by a large factor, because each pixel needs one bit instead of eight or more. The saving depends on how the image is stored afterwards.
Can I undo binarization?
No. Once the grey tones are thrown away they cannot be rebuilt, so keep the original scan if you may need the faint details later.
Is binarization the same as OCR?
No. Binarization only changes the colors of pixels, while OCR reads the shapes and produces characters. The first is often a preparation step for the second.

Side-by-side comparisons

Related terms

Need more than a quick fix?

AuraDoc is a document platform with OCR, AI search and summaries, and more than 60 PDF tools for teams. Try it free for 7 days.

Free tools for personal and professional use. This one runs on an isolated AuraDoc server: the file is deleted right after processing and never stored. Fair use and anti-abuse policy