ocrmypdf

TLDR

Add OCR layer to PDF

$ ocrmypdf [input.pdf] [output.pdf]

OCR and correct skew

$ ocrmypdf --deskew [input.pdf] [output.pdf]

OCR and clean background

$ ocrmypdf --clean [input.pdf] [output.pdf]

Specify language

$ ocrmypdf -l [deu] [input.pdf] [output.pdf]

Multiple languages

$ ocrmypdf -l [eng+fra] [input.pdf] [output.pdf]

Force OCR even if text exists

$ ocrmypdf --force-ocr [input.pdf] [output.pdf]

Skip pages with text

$ ocrmypdf --skip-text [input.pdf] [output.pdf]

Optimize and reduce size

$ ocrmypdf --optimize [3] [input.pdf] [output.pdf]

SYNOPSIS

ocrmypdf [-l lang] [--deskew] [--clean] [--force-ocr] [--optimize level] [options] input.pdf output.pdf

ocrmypdf adds an OCR text layer to scanned PDFs, making them searchable and selectable. It uses Tesseract OCR and outputs PDF/A for archival quality by default.
The tool preserves the original visual appearance while adding invisible text behind the scanned images. This means the file looks identical but text can be copied, searched, and indexed.
Image preprocessing improves OCR accuracy: deskew corrects tilted scans, clean removes noise and artifacts, and rotate-pages fixes orientation. These can significantly improve results on poor-quality scans.
Multiple languages can be combined (eng+fra+deu). Language packs must be installed for Tesseract. The tool detects existing text to avoid double-processing unless forced.
Optimization levels reduce file size through image recompression. Level 3 uses aggressive JBIG2 compression suitable for archival. PDF/A output ensures long-term readability.
Parallel processing speeds up multi-page documents. Progress is shown by default. Sidecar output extracts just the text for external processing.

PARAMETERS

-l LANG, --language LANG

OCR language (Tesseract language codes).

--deskew

Correct page skew before OCR.

--clean

Clean page background before OCR.

--clean-final

Clean and keep cleaned image in output.

--rotate-pages

Rotate pages to correct orientation.

--remove-background

Remove background from pages.

--force-ocr

OCR all pages, replacing existing text.

--skip-text

Skip pages that already have text.

--redo-ocr

Redo OCR on pages with existing text.

--optimize LEVEL

Optimize output (0=off, 1-3 increasing).

--output-type TYPE

Output type: pdf, pdfa, pdfa-1, pdfa-2, pdfa-3.

--pdfa-image-compression TYPE

Compression: jpeg, lossless.

-j NUM, --jobs NUM

Number of parallel jobs.

--image-dpi DPI

DPI for images without metadata.

-q, --quiet

Suppress output.

-v, --verbose [LEVEL]

Verbose output (0-2).

--sidecar FILE

Write OCR text to sidecar file.

CAVEATS

OCR accuracy depends on scan quality. Very low resolution or heavily compressed images may produce poor results. Language packs must be installed separately. Some complex layouts may not OCR well. PDF/A conversion may strip some features. Processing large PDFs requires significant memory.

HISTORY

ocrmypdf was created by James R. Barlow starting around 2013. It wraps Tesseract OCR with intelligent PDF handling, image preprocessing, and PDF/A output. The project addressed the need for a complete solution to make scanned PDFs searchable, automating what was previously a manual multi-step process.