Methodology for OCR error checking through text image regeneration

• US 5,889,897 A
• Filed: 04/08/1997
• Issued: 03/30/1999
• Est. Priority Date: 04/08/1997
• Status: Expired due to Term

• Current Assignee: Hanger Solutions, LLC (IP Investments Group LLC)
• Original Assignee: International Patent Holdings Limited
• Inventors: Medina, Mitchell
• Primary Examiner(s): SHALWALA, BIPIN H
• Application Number:
  US08/835,540
• Time in Patent Office: 721 Days
• Field of Search: 382/309, 382/310, 382/311, 382/181, 382/182, 382/218, 382/219, 382/220, 382/227, 382/228, 364/737, 371/3, 707/530, 707/531
• US Class Current: 382/310
• CPC Class Codes: G06V 10/98 Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns

Abstract

A method of reducing errors in optical character recognition (OCR) procedures includes digitizing an image of an object, the image containing alpha-numeric characters, and storing a bitmap of the digitized image as a scanned document file (SDF). An OCR step is performed to obtain at least one candidate character, and an indication of the at least one candidate character is stored in a textural results file (TRF). The font of the digitized image is determined and the font is stored in a regeneration library file (RLF). A regenerated image file is generated using the TRF and the RLF. At least a portion of the regenerated image file is compared with a corresponding portion of the bitmap of the digitized image stored in the SDF. The TRF is outputted if the results of the comparison indicate a match of at least the portion of the regenerated image file with the corresponding portion of the bitmap. If a match is not found, further processing is performed to resolve the mismatch.

Claims

1. A method of reducing errors in optical character recognition procedures comprising the steps of:

(a) digitizing an image of an object, said image containing alpha-numeric characters;

(b) storing a bitmap of said digitized image as a scanned document file (SDF);

(c) performing an OCR step to obtain at least one candidate character;

(d) storing an indication of said at least one candidate character in a textural results file (TRF);

(e) determining the font of said digitized image;

(f) storing said determined font in a regeneration library file (RLF);

(g) generating a regenerated image file using said TRF and said RLF;

(h) comparing at least a portion of said regenerated image file with a corresponding portion of the bitmap of said digitized image stored in said scanned document file;

(i) outputting said TRF if the results of said comparison step indicate a match of at least said portion of said regenerated image file with said corresponding portion of said bitmap in said scanned document; and

(j) performing further processing to resolve the mismatch if a match is not found in step (i).

2. The method as recited in claim 1 further including the step of:

determining at least one placement characteristic of said digitized image, and wherein step (h) includes the step of utilizing the placement characteristic to determine the correspondence of said portion of said regenerated image file with the portion of said scanned document file.

3. The method as recited in claim 2 wherein step (j) includes:

providing a predetermined output indication at a position corresponding to a portion of said digitized image which does not match said corresponding portion of said regenerated image file.

4. The method as recited in claim 3 wherein said further processing includes operator checking of said digitized image at a position corresponding to said predetermined output indication.

5. The method as recited in claim 1 wherein step (j) includes:

providing a predetermined output indication at a position corresponding to a portion of said digitized image which does not match said corresponding portion of said regenerated image file.

6. The method as recited in claim 5 wherein said further processing includes operator checking of said digitized image at a position corresponding to said predetermined output indication.

7. The method as recited in claim 1 in which step (j) includes the use of at least one of the following machine resolution techniques:

a) utilizing alternative OCR steps from those performed in step (c);

b) utilizing additional OCR steps from those performed in step (c);

c) utilizing voting among multiple OCR steps;

d) utilizing alternative regeneration step from utilized in step (g); and

e) utilizing an initial comparison threshold in the comparing step (h) and utilizing an alternative comparison threshold from the initial comparison threshold used in step (h).

8. The method as recited in claim 7 in which step (j) includes:

human arbitration of any mismatches which are not resolved by one or more of said machine resolution techniques.

9. The method as recited in claim 8 wherein step (j) includes:

providing a predetermined output indication at a position corresponding to a portion of said digitized image which does not match said corresponding portion of said regenerated image file.

10. The method as recited in claim 9 wherein said further processing includes operator checking of said digitized image at a position corresponding to said predetermined output indication.

11. The method as recited in claim 2 in which step (j) includes the use of at least one of the following machine resolution techniques:

a) utilizing alternative OCR steps from those performed in step (c);

b) utilizing additional OCR steps from those performed in step (c);

c) utilizing voting among multiple OCR steps;

d) utilizing alternative regeneration step from utilized in step (g); and

e) utilizing an initial comparison threshold in the comparing step (h) and utilizing an alternative comparison threshold from the initial comparison threshold used in step (h).

12. The method as recited in claim 11 in which step (j) includes:

human arbitration of any mismatches which are not resolved by one or more of said machine resolution techniques.

13. The method as recited in claim 12 wherein step (j) includes:

providing a predetermined output indication at a position corresponding to a portion of said digitized image which does not match said corresponding portion of said regenerated image file.

14. The method as recited in claim 13 wherein said further processing includes operator checking of said digitized image at a position corresponding to said predetermined output indication.