Systems for recognizing printed characters

Abstract

A system and method are disclosed for character recognition of characters sposed upon a document. Initially, a set of characters on the document is read to obtain corresponding information data, and their associated error codes and the data are stored in a first of N zones of a memory of the system. If an examination of the error codes so stored indicates that at least one character is erroneous, the set of the characters of the document is re-read under differing reading conditions and the new corresponding information data and associated error codes are derived and stored in different memory zones of the memory for as long as at least one erroneous character is detected. A correcting sequence is initiated when a number N of reading operations corresponding to the number of memory zones has been performed and at least one character is recognized as being erroneous; the correcting sequence includes checking the error code associated with the information data read from the first address of the first memory zone of the memory to determine whether there has been an erroneously-read character. If an error is detected, the readout of data from the same address in the next memory zone is effected, while if an error has not been detected, a readout of information data from the next address in the same memory zone is effected. If no error is detected, the information data is read out and stored in a data buffer, while, on the other hand, the document is rejected if the information data as derived from the last address of any one of the N memory zones of the memory proves to be erroneous.