OCR = Optical Character Recognition (interpreting machine print)
ICR = Intelligent Character Recognition (interpreting handwriting)
Tests for numbers and marks can be added (interval check, max number of marks, checksums, table lookups and much more). A verification dialog is used to correct unrecognised letters and errors detected by other tests.
Scanning speed range from 1000 to over 7000 double sided sheets pr hour, depending on the scanner model.
The interpreted data and the scanned images may be exported to files, databases or electronic archives.
Since 1994 I have developed OCR & ICR solutions for everything from Medical lab forms, Parking tickets, Electricity usage forms, and of course the norwegian election (scanning and ICR interpreting election ballots).
There is also OCR software for scanning and interpreting invoices. It looks for moving data by interpreting all text on the invoices the first time, looking for the «intro text» in front of the fields, remembering the locations for later.
More about the election scanning system: EVRY
International info on the ocr/icr engine: ReadSoft
Below are some screenshots from the election scanning:
«Drop-out» colors that the scanner can't see is a good way
to prevent «boxes» from interferring with interpretation
Verification of power usage form with uncertain digits
If a form cannot be identified during scanning it can be
manually identified right there or removed and rescanned
In norwegian elections you can add names from other lists.
Verification shows a list with the best ICR match first
