This is the question we hear most often: I scanned a document, opened the PDF, but I can’t select any text and search finds nothing. The file looks like text but behaves like a picture. And that is exactly what it is – a picture.
Two completely different kinds of PDF
The same name, „PDF”, covers two different things.
A digital PDF is created on a computer – saved from Word, a printer driver or another application. It contains real text: characters, fonts, lines. In a file like this, search works and you can select and copy text.
A scanned PDF is a photo of paper wrapped in a PDF container. As far as the computer is concerned, there are no letters in it at all – just a collection of dots. That is why search finds nothing: there is simply nothing to search.
This is where OCR comes in
OCR (optical character recognition) is the technology that recognises characters in an image and turns them into text. The software looks at a cluster of dots and decides: this is the letter „a”, this is „á”, and this is a table border.
The result is usually saved in one of two ways.
Searchable PDF. The document looks exactly like the scan – the same image, the same stamps and handwritten notes – but an invisible text layer is hidden underneath the image. The document looks like the original, yet search works. For archives, this is almost always the right choice.
Editable file. The scan is converted into a Word, Excel or other file in which the text is real and can be edited. The software tries to rebuild not only the characters but also the layout: columns, tables, headings.
Why recognition is excellent in one case and poor in another
Recognition quality doesn’t depend on the software alone. In practice, a few things make the difference.
- Scan resolution. 300 dpi is the minimum for a decent result. A 150 dpi scan looks readable to a person, but for the software small characters blur together.
- Skew and perspective. A page photographed with a phone is almost always at an angle. Good software corrects this automatically – FineReader has image pre-processing tools that remove the background and straighten the perspective before recognition.
- Language. If the correct language isn’t selected, letters with diacritics can turn into incorrect characters. FineReader recognises 198 languages and, more importantly, lets you combine them – a document can contain, say, Lithuanian and English at the same time.
- Type of lettering. Printed text is recognised well. Handwriting isn’t, and it’s not worth expecting it to be.
Do you need to check the result?
Yes, but not all of it. Decent software shows you which places it was unsure about – FineReader’s verification tool shows the characters that may have been recognised incorrectly. Reviewing ten doubtful spots is far quicker than reading the whole document.
Tables and figures are a separate story. If you scan invoices or reports where every digit matters, check the totals. Letters are rarely misrecognised, but „0” and „O” or „1” and „l” are classic examples of easily confused characters.
A practical benefit that often goes unnoticed
Once an archive becomes searchable, the way you work changes. You no longer need to remember which folder holds what – just type a surname or a contract number. Documents that nobody can find effectively don’t exist, even if they are physically stored.
The second benefit is file size. Scans take up a lot of space, but after processing with compression the same document can shrink several times over and still remain readable.
What to do now
If you have an archive of scans you can’t search, choose the worst-quality document and test it. Not a demo file from a website, but the poor-quality scan that caused problems in the first place. You can download the trial version free of charge, and the result in ten minutes will tell you more than any description.
Once the text has been recognised, it can also be edited – we explain how in our article how to edit a scanned document.
Text recognition (OCR) is included in every ABBYY edition: FineReader PDF Standard and FineReader PDF Corporate for Windows computers, and FineReader PDF for Mac for Mac. You can buy a licence in our shop; the key is sent by email.