ocrmypdf 命令

常用示例

说明

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.

参数

-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.