Editing scanned PDFs with OCR

A scanned PDF is a picture of a page. There is no text inside it to select, search or edit — only pixels. blupdf runs optical character recognition (OCR) on every scanned page so the words become real, editable blocks, then treats the page exactly like a digital PDF.

What counts as a scan

blupdf classifies each page when the document is uploaded. A page is treated as scanned when it contains no usable text layer — typical for output from a scanner or a multifunction printer, a photo taken with a phone, a fax, or a PDF that was "printed" to an image. Mixed documents are handled page by page: a contract with two scanned signature pages and ten digital pages costs OCR only for the two scanned ones.

JPG and PNG images can be uploaded directly; blupdf wraps each image into a PDF page before OCR.

How OCR runs on upload

When a scanned page is detected, blupdf sends the page image to its AI OCR model, which returns the text together with the position, size and reading order of every block. The result is stored with the document, so OCR runs once per page — not every time you open the editor. The charge is 80 credits per scanned page (see pricing).

Accuracy and what affects it

Modern AI OCR is very accurate on clean office documents and noticeably less so on poor input. The factors that matter most:

• Resolution. 300 dpi is ideal. Below 150 dpi, small text and punctuation start to blur together.
• Skew and perspective. A page photographed at an angle stretches letters unevenly; scan flat when you can.
• Lighting and shadows. Phone photos with a shadow across the page lose characters in the dark band.
• Handwriting. Printed text is reliable; handwriting is recognized best-effort and should be checked line by line.
• Language. blupdf's OCR handles most Latin-script languages and common non-Latin scripts; mixed-language pages are fine.

Font estimation on scans

A scan carries no font information, so blupdf estimates the typeface from the glyph shapes: serif or sans, weight, width, x-height and letter spacing. It then picks the closest match from its font library and uses that font for your new text. The result is designed to blend with the surrounding words at reading size; it will not be the identical font, because the identical font is not in the file. The block toolbar shows the chosen font and lets you switch to another match.

Fixing OCR mistakes

Because OCR output is just text in blocks, correcting a misread is the same as any edit: click the block and fix the characters. Common misreads are 0/O, 1/l/I, and rn/m. If a whole region was missed — for example a stamp or a rotated label — you can add a text block by hand and position it over the area.

Exporting a searchable PDF

When you export, blupdf writes the recognized and edited text into the PDF as a real text layer on top of the original page image. The exported file is searchable, its text can be copied, and screen readers can read it. Blocks you edited are rendered in the matched font; blocks you did not touch keep the original scanned appearance with an invisible text layer behind them.

Photos and images as input

Photos work the same way as scans, with the extra step of being wrapped into a page. For best results, crop to the page, keep the camera parallel to the paper, and avoid flash glare. See Adding images and signatures if you want to place an image into a document rather than edit one.