Turn an image-only scan into a searchable PDF
A scanner often saves every page as a picture. The words look normal, but search, copy and accessibility tools cannot read them. FigureOwl uses optical character recognition (OCR) to identify English words and places an invisible text layer over the original page. The scan remains visible while Ctrl+F, text selection and PDF indexing can use the recognized text.
Keep the original page appearance
The downloaded file preserves the source PDF page as its visible content. OCR words are positioned behind the scan rather than replacing it with newly typeset text. This is useful for signed forms, invoices, receipts, archived letters and other documents where the original appearance matters. Always compare important names, dates, totals and reference numbers against the scan because OCR can make mistakes.
Choose the recognition quality
Fast mode renders pages at 150 DPI and suits clean printed pages. Balanced mode uses 200 DPI and is a sensible default. Best mode uses up to 300 DPI for small or less distinct type, but needs more memory and takes longer. Very large pages are capped to protect the browser from excessive memory use.
Handle mixed digital and scanned PDFs
Some PDFs combine normal digital pages with image-only scans. Keep “Skip pages that already contain text” enabled to preserve the existing selectable text and run OCR only on pages that need it. Turn the option off only when a page has an unusable or incomplete text layer and you intentionally want another recognition layer.
Download recognized text when useful
The optional TXT file contains the recognized words page by page. It is handy for a quick text review, searching outside a PDF reader or moving a rough draft into another application. OCR does not reproduce columns, tables, typography or reading order perfectly, so treat the text file as extracted content rather than a faithful document layout.
Know the limits of OCR
Clear, upright, high-contrast printed English gives the best result. Blur, shadows, compression artifacts, decorative fonts, skewed pages, handwriting and text over patterned backgrounds reduce accuracy. This tool currently recognizes English only. It does not translate text, verify facts, interpret signatures or certify that a copy is legally accurate.
Private processing in your browser
Your selected PDF is processed on your device and is not uploaded to FigureOwl. The OCR engine and English language model are downloaded when recognition starts and may be cached by the browser. Because OCR is demanding, each run is limited to one PDF of 30 MB or less and no more than 20 pages.
Review sensitive and official documents
Searchability is convenient, but the invisible layer can contain recognition errors that are not obvious from the visible scan. Check extracted text before relying on it for contracts, tax records, medical information, legal filings, account numbers or accessibility. The new copy does not preserve certificate signatures, comments, bookmarks, attachments or every interactive PDF feature.