Systems and methods for capturing critical fields from a mobile image of a credit card bill

• US 10,509,958 B2
• Filed: 10/28/2016
• Issued: 12/17/2019
• Est. Priority Date: 03/15/2013
• Status: Active Grant

• Current Assignee: Mitek Systems Incorporated
• Original Assignee: Mitek Systems Incorporated
• Inventors: Kliatskine, Vitali, Nepomniachtchi, Grigori, Kotovich, Nikolay
• Primary Examiner(s): Iwarere, Oluseye
• Application Number:
  US15/338,203
• Publication Number:
  US 20170109574A1
• Time in Patent Office: 1,145 Days
• Field of Search: 382181, 3404255, 705 34
• CPC Class Codes: G06K 7/10712 Fixed beam scanning
• CPC Class Codes: G06K 7/1413 1D bar codes
• CPC Class Codes: G06Q 30/04 Billing or invoicing

Abstract

The embodiments herein focus on improving recognition accuracy of these fields on credit card bills by detecting and identifying critical fields on a credit card, extracting the data from the critical fields and comparing the data with known data on a payor, payee and biller.

Claims

1. A method for identifying contents of a credit card bill from an image of the credit card bill, comprising:

receiving the image of the credit card bill which includes a codeline field;

identifying, from the image of the credit card bill, at least one field of the credit card bill using at least one of a keyword-based search and a format-based search, wherein the at least one field is not the codeline field;

extracting, from the received image, content of the at least one identified field;

capturing codeline content of the codeline field;

removing nonessential punctuation marks from the captured codeline content;

cross-validating the extracted content of the at least one identified field against the captured codeline content to determine if the extracted content matches a portion of the captured codeline content; and

when the extracted content matches,

replacing the extracted content with the matching portion of the captured codeline content as the identified content of the at least one identified field of the credit card bill; and

outputting the identified content of the at least one identified field of the credit card bill.

2. The method of claim 1, wherein the keyword-based search identifies the at least one field by identifying a predetermined keyword located near the at least one field.

3. The method of claim 1, wherein the format-based search identifies the at least one field by matching a format of the at least one field with a predetermined format of the at least one field.

4. The method of claim 1, wherein the at least one identified field is the Account Number field which includes an account number associated with the credit card bill.

5. The method of claim 4, wherein the codeline field includes the account number associated with the credit card bill.

6. The method of claim 1, wherein the at least one identified field is the Balance Due field which includes an amount due associated with the credit card bill.

7. The method of claim 6, wherein the codeline field includes the amount due associated with the credit card bill.

8. The method of claim 1, wherein the portion of the captured codeline content is a substring of the captured codeline content.

9. The method of claim 1, wherein prior to determining if the extracted content matches a portion of the captured codeline content, the method further comprises removing spaces, dashes and other nonessential punctuation marks from both the capturing codeline content and the portion of the captured codeline content.

10. The method of claim 1, wherein determining if the extracted content matches a portion of the captured codeline content includes using a matching threshold which allows at most a single-character difference between the extracted content matches a portion of the captured codeline content.

11. The method of claim 1, wherein determining if the extracted content matches a portion of the captured codeline content includes using a matching threshold which allows both a single-character difference and additionally differences involving frequently-misrecognized characters between the extracted content and the portion of the captured codeline content.

12. The method of claim 1, wherein the method further comprises computing a confidence score for the extracted content of the at least one identified field based on a combination of a confidence of the keyword-based search and/or format-based search and whether the extracted content matches a portion of the captured codeline content.

13. The method of claim 12, wherein the confidence score is further computed by determining a geometrical alignment score reflecting the location of a keyword of the keyword-based search relative to the extracted content of the at least one identified field.

14. The method of claim 12, wherein the confidence score increases if a portion of the captured content of the codeline matches the extracted content of the at least one identified field.

15. The method of claim 14, wherein the confidence score depends on a degree of matching between the portion of the captured content of the codeline and the extracted content of the at least one identified field.

16. The method of claim 1, wherein the matching portion of the captured codeline content has a higher accuracy than the extracted content of the at least one identified field.