ABBYY uses MRC for their multifunction printers.
http://www.abbyy.com/ocr_sdk_embedded/for_MFP_vendors/
And the ABBYY software incorporates MRC for PDFs.
They say this.
“The MRC technology analyzes the outlines of similar characters in the document, creates an average character template and uses it instead of a character itself. This leads to better readability, because some of the text defects are corrected, and the character outlines become more precise.”
http://www.abbyy.com/ocr_sdk_embedded/pdf_mrc/
Sounds like it creates pixel for pixel identical characters.
“The MRC technology analyzes the outlines of similar characters in the document, creates an average character template and uses it instead of a character itself.”
This makes it sound like MRC is OCR, but it is not OCR character identification. Instead of ‘character” I prefer shape detection and replication or butterdezillion’s “blob” detection and replication...
Any imaginable form, shape, image, blob (or character or number) can be detected and cataloged as a scan progresses and then when a “close enough” similar shape is encountered, the first one is used instead at that bitmap location in the scan. This reduces (compresses) the size of the scanned document pdf file.