Post-processing: Automated OCR and PDF creation

Owned by MM Hanke
Last updated: Dec 06, 2024
2 min read

Prerequisites

Software

Make sure the versions of tesseract, muPDF and GhostScript are up to date. Run the following command on your command line to update software components of your Linux installation:

sudo apt update && sudo apt upgrade

Input files and staging space

This process presumes that the image files used to run OCR will be cropped and QC'd TIF files, named in sequential order. Avoid using this method on JPEG files, as this could result in significant loss of image quality.

Page images need to be staged in one subdirectory per volume/issue.

Ensure that the staging area used to process the images has enough space available. The process will create intermediary files (page-level PDFs and an uncompressed/unoptimized aggregate PDF document), which can take up hundreds of megabytes.

Processing instructions

Doesn't work as described here, still missing instructions to place the PDF/A instructions and ICC profile files: High Level Devices — Ghostscript 10.05.0 documentation

The path under step 4 (using Ghostscript) needs to be valid for the OS/computer from which you are running the command. The PS file the path points to needs to be edited to contain a valid path pointing to an ICC profile.

  1. Navigate to the staging directory containing the Production Master (_pm) image files

  2. Create temporary subdirectories 'pdf' and 'pdf/pages', and navigate back to the staging directory:

    mkdir pdf cd pdf mkdir pages cd ..

    Run tesseract OCR on all images in the staging directory: 

    for i in *.tif; do tesseract -c tessedit_page_number=0 -l eng $i pdf/pages/${i%_pm.tif} pdf; done

  3. Merge page-level PDF documents into one PDF per volume/issue:

  4. Produce the final PDF document.

  5. Delete the pdf subfolder and combined-pdf.pdf

 

 

 

Confluence Documentation | Web Privacy Policy | Web Accessibility