Searchable PDF: image versus document

If you scanned a sheet, saved it as PDF, opened it and found it identical to the original — that does not mean you have archived anything. What you most likely have is an image placed inside a PDF file, and that is a completely different thing from a searchable document.

The five-second test

Open any file from your archive, press Ctrl+F (or Command+F on a Mac) and type a word you can see in front of you in the document. If the file finds it, it is searchable. If nothing comes up, it is an image. Run that test on a sample of your existing archive right now — the result surprises a lot of…

What actually happens inside

The scanner photographs the sheet. To turn that image into text, OCR runs over it — reading the shape of each character and converting it into an actual character. The result is placed in a hidden text layer behind the image: you see the original exactly as it was, and the computer sees the text. Th…

Arabic and OCR

Recognising Arabic characters is harder than English for reasons rooted in the script: letters connect, a letter's shape changes with its position in the word, and dots distinguish between many characters. Modern software has become very good with printed Arabic — accuracy on a clean printed documen…

What ruins the result

• A faint document, or a copy of a copy A sheet fed at an angle — characters get misread Scanning resolution too low (under 300 dpi for text) Coloured backgrounds or stamps over the text Handwriting or decorative fonts

The right resolution

300 dpi is the practical standard for text documents — below that OCR starts making mistakes, and above it the file grows without a real gain in accuracy. Photographs and engineering drawings are a different matter and need more. But for invoices, contracts and delivery notes, 300 is the number.