Abstract: A method for scanning and sorting documents which employs data not found on the document to make the sorting decision. Provision is also made for correcting misread information with the most probable correct data so that the scanning and sorting process may continue uninterrupted. In one form, the method comprises the steps of: (a) reading first data from a document by optical character recognition; (b) accessing second data which contains a list of valid first data; (c) determining if said data contains an error based on said second data; (d) accessing third data if said first data contains an error, said third data containing a probable correction of said first data; (e) changing said first data to equate with said third data; (f) retrieving fourth data from a data file based upon said first; (g) sorting said document based upon said fourth data.

Abstract: A redundant character recognition system which minimizes both read failures and ambiguities caused by conflicting read signals by merging character recognition signals from MICR, OCR and Bar-Code read heads to produce a single data stream for automatic sorting of a constant velocity train of documents. In the event the MICR and OCR readers are unable to identify a character image, a Bar-Code reader is used to supplement MICR/OCR data by replacing MICR data field rejects with bar-code data field recognition signals. When rejects are generated from both the MICR and bar-code data fields, only self-checking fields within the bar-code data field are merged upon a successful self-check test being performed. If self-checking errors occur, the MICR read data is restored.