Hierarchical character recognition system

• US 5,881,172 A
• Filed: 12/09/1996
• Issued: 03/09/1999
• Est. Priority Date: 12/09/1996
• Status: Expired due to Term

• Current Assignee: Mitek Systems Incorporated
• Original Assignee: Mitek Systems Incorporated
• Inventors: Pintsov, David A.
• Primary Examiner(s): Tran, Phuoc
• Application Number:
  US08/761,845
• Time in Patent Office: 820 Days
• Field of Search: 382/224, 382/226, 382/227, 382/229, 382/231, 382/310, 382/321, 382/187
• US Class Current: 382/227
• CPC Class Codes: G06V 30/2504 Coarse or fine approaches, e.g. resolution of ambiguities or multiscale approaches

Abstract

A method and apparatus for hierarchical character recognition processing of ambiguous and noisy characters which produces highly reliable results at high levels of hierarchical processing. The invention first applies a universal classifier system (which may comprise one or more universal classifiers) to input image data, and identifies "suspicious" characters. The image data for suspicious characters is then applied to a "specialist" classifier that is designed to handle only a narrow and well-defined set of recognition cases. This hierarchical processing architecture and method results in increased accuracy of recognition. The method is particularly applicable to handwritten characters and to distorted and noisy machine-printed characters.

Claims

1. A hierarchical recognition system including:

(a) a universal classifier system for recognizing a probable character from input image data, for determining if the probable character is a suspicious character, and for outputting a character code for each non-suspicious character;

(b) at least one specialist classifier, each for recognizing a distinct ambiguity class of characters, and selectably coupled to the universal classifier system;

(c) means for selecting and applying a specialist classifier corresponding to each suspicious character, whereby the selected specialist classifier outputs a character code for each suspicious character.

2. A method for recognizing characters from input image data, comprising the steps of:

(a) recognizing a probable character from the input image data;

(b) determining if the probable character is a suspicious character;

(c) outputting a character code for each non-suspicious character, and otherwise selecting a specialist classifier, configured to recognize a distinct ambiguity class of characters, corresponding to each suspicious character;

(d) recognizing each suspicious character with the selected specialist classifier; and

(e) outputting a character code for each suspicious character.

3. A computer program, residing on a computer-readable medium, for recognizing characters from input image data, the computer program comprising instructions for causing a computer to:

(a) recognize a probable character from the input image data;

(b) determine if the probable character is a suspicious character;

(c) output a character code for each non-suspicious character, and otherwise select a specialist classifier, configured to recognize a distinct ambiguity class of characters, corresponding to each suspicious character;

(d) recognize each suspicious character with the selected specialist classifier; and

(e) output a character code for each suspicious character.