Lab: turn a stack of scans into a searchable, queryable archive
Scenario
A relative gives you a folder: ten phone photos of old letters (some sideways, one crooked), plus a scanned PDF of a typed report. You want one archive PDF where you can find any name, a plain-text copy of the typed report, and a check of who created the report file.
What you will learn
Photos to PDF, orientation fixing before recognition, Searchable with language and deskew, Search with a highlighted result, Extract Text, Extract Metadata, and when to use OCR Image instead.
Steps
- Rotate. Put the letter photos into Merger in date order (Sequential mode) to get one PDF. Open it and use Rotation on the sideways pages.
- Make it searchable. Run Searchable on this PDF: set the correct document language, keep Deskew on, tick Remove background/watermark if the paper is stained. Download the result.
- Test it. Open the PDF, press Ctrl+F and look for a name you can see on a page.
- Search. Run Search on the searchable PDF for a family name. Open the result and jump through the "Search results" bookmarks.
- Report text. Run Searchable on the scanned report, then Extract Text (Pure mode) on its result. Skim the .txt for errors.
- Who made it? Run Extract Metadata on both PDFs and compare producer and creation date.
- Quick alternative. Upload one letter photo to OCR Image and compare its .txt with the hidden text from step 2.
- Shrink. Run Compressor on the final archive PDF.
Checkpoints
- Every page is upright before step 2.
- Ctrl+F finds a visible word; if not, check language and rotation.
- Search returns a marked PDF with highlighted hits (or tells you nothing was found).
- The .txt from the report has the right words, with a few recognition errors at most.
- Metadata shows a producer or creation date for each file (or empty fields if none were recorded).
Going further
Add Add TOC to give the archive chapters, use Ask PDF to ask questions about the letters, and see the digitise an archive recipe.