Glossary · OCR (optical character recognition)
What is OCR (optical character recognition)?
Definition: OCR is the technology that converts text in an image, such as a scan, a photo or a PDF without a text layer, into digital text a computer can process.
OCR analyses the image, locates the areas containing text and recognises each character to produce digital text. Current systems use machine learning models and cope well with different typefaces, languages and a degree of skew or noise. Some also recognise handwriting, though with lower accuracy. Image quality remains the factor with the greatest influence on the result.
In a business, it is the first step in working with paper or scanned documents: invoices, signed delivery notes, old contracts or identity documents. It makes archives that used to be mere images searchable and feeds data extraction processes. On its own, however, it delivers plain text, without knowing which number is the total and which is the postcode.
That is why OCR is rarely enough to automate a process. Intelligent document processing adds the layer that interprets the content: it classifies the document, identifies each field and validates it. Templates based only on fixed positions fail when a sender changes its layout, something AI models handle better. Many PDFs already contain text and do not need OCR at all.
Example: signed delivery notes that drivers photograph with their phones are run through OCR so the system can read the delivery note number and file it with the matching order.