Confusion set based method and system for correcting misrecognized words appearing in documents generated by an optical character recognition technique

An apparatus for recognizing at least one word in a document, the word including at least one predetermined character member, the apparatus comprising:

• a) first means for providing a recognized word based on the word in the document;
• b) first means for determining whether the recognized word has been misrecognized;
• c) second means for providing, if the recognized word has been misrecognized, a set of reference words, each reference word comprising a different set of predetermined character members;
• d) third means for providing a plurality of confusion sets, each confusion set grouping together a different plurality of character members, a content of each confusion set being independent of the recognized word;
• e) means for comparing at least one character sequence of the misrecognized word with a corresponding character sequence of a current one of the set of reference words to determine which corresponding character sequences do not include the same character members;
• f) first means for eliminating any one of the set of reference words if the character member of any character sequence of the misrecognized word does not correspond to the character member of the corresponding character sequence in the reference word and if the character members from the corresponding character sequences in the misrecognized word and the reference word are not from the same confusion set, the remaining non-eliminated reference words comprising a set of candidate reference words;
• and
• g) means for selecting one of the set of candidate reference words in accordance with a set of predetermined criteria, the select candidate reference word comprising a replacement word for the misrecognized word.

An apparatus for recognizing at least one word in a document, the word including at least one predetermined character, the apparatus comprising:

• a scanning device;
• and
• a processing device in communication with the scanning device, wherein the processing device comprises:
• a central processing unit,
• an optical character recognition module in communication with the centrol processing unit,
• a confusion matrix memory in communication with the central processing unit,
• a word checking module in communication with the central processing unit
• a confusion set generating module in communication with the central processing unit, each confusion set having a content that is independent of an input to the scanning device,
• a confusion set memory in communication with the central processing unit, and
• a word correction module in communication with the central processing unit.

An apparatus, for recognizing at least one word in a document, the word including at least one predetermined character, the apparatus comprising:

• a scanning device;
• and
• a processing device in communication with the scanning device, wherein the processing device comprises:
• a central processing unit,
• an optical character recognition unit in communication with the central processing unit,
• a word checking module in communication with the central processing unit,
• a confusion set memory in communication with the central processing unit and provided with a plurality of predetermined confusion sets, each predetermined confusion set having a content that is independent of an input to the scanning device, and
• a word correction module in communication with the central processing unit.

A method of recognizing at least one word in a document, the word including at least one predetermined character member, the method comprising the steps of:

• a) providing a recognized word based on the word in the document;
• b) determining whether the recognized word has been misrecognized;
• c) providing, if the recognized word has been misrecognized, a set of reference words, each reference word comprising a different set of predetermined character members;
• d) providing a plurality of confusion sets, each confusion set grouping together a different plurality of character members;
• e) comparing at least one character sequence of the misrecognized word with a corresponding character sequence of a current one of the set of reference words to determine which corresponding character sequences do not include the same character members;
• f) eliminating the current reference word if the character member of any character sequence of the misrecognized word does not correspond to the character member of the corresponding character sequence in the current reference word and if the character members from the corresponding character sequences in the misrecognized word and the current reference word are not from the same confusion set;
• g) repeating steps e) and f) for each reference word of the set of reference words, the remaining non-eliminated reference words comprising a set of candidate reference words;
• and
• h) selecting one of the set of candidate reference words in accordance with a set of predetermined criteria, the selected candidate reference word comprising a replacement word for the misrecognized word, wherein the confusion sets are derived from at least one confusion matrix, and wherein the at least one confusion matrix is generated prior to deriving the plurality of confusion sets therefrom.

An apparatus for recognizing at least one word in a document, the word including at least one predetermined character member, the apparatus comprising:

• a) first means for providing a recognized word based on the word in the document;
• b) first means for determining whether the recognized word has been misrecognized;
• c) second means for providing, if the recognized word has been misrecognized, a set of reference words, each reference word comprising a different set of predetermined character members;
• d) third means for providing a plurality of confusion sets, each confusion set grouping together a different plurality of character members;
• e) means for comparing at least one character sequence of the misrecognized word with a corresponding character sequence of a current one of the set of reference words to determine which corresponding character sequences do not include the same character members;
• f) first means for eliminating any one of the set of reference words if the character member of any character sequence of the misrecognized word does not correspond to the character member of the corresponding character sequence in the reference word and if the character members from the corresponding character sequences in the misrecognized word and the reference word are not from the same confusion set, the remaining non-eliminated reference words comprising a set of candidate reference words;
• and
• g) means for selecting one of the set of candidate reference words in accordance with a set of predetermined criteria, the select candidate reference word comprising a replacement word for the misrecognized word, and wherein the at least one confusion matrix is generated prior to deriving the plurality of confusion sets therefrom.