OCR stands for Optical Character Recognition.
When you scan a document, the scanner passes over the document and produces a representation of it on the computer using pixels, or in other words, very little dots. It does not consider text any differently than other shapes that it sees. As a result, you have a document that does not have any actual words or text.
This poses various issues such as you cannot text edit, select, copy and paste and all the wonderful things that you can do with text. To deal with this, there is OCR software which essentially looks at the shapes and interprets them into a text format. It converts the "image" into a text document.
Scanning and OCR are often done in two separate steps as OCR software is often not included with the scanner. Some scanners do combine the two though, such as the ScanSnap scanner in the JCU Library.