The acronym OCR stands for “Optical Character Recognition”. OCR scans a document and, by means of learned alphabet or semiotics (e.g. Latin, Russian, Kanji, Hiragana …) determines the letter matching the scanned character.
The text thus read is stored as additional so-called “full text information” to the document, making the entire document searchable.
However, it is important to note that not every OCR works equally well, for it always depends on how well an existing OCR has been or can be trained.