Personal information retrieval using knowledge bases for optical character recognition correction

• US 7,826,665 B2
• Filed: 12/12/2005
• Issued: 11/02/2010
• Est. Priority Date: 12/12/2005
• Status: Active Grant

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Bressan, Marco, Dance, Christopher R., Dejean, Herve
• Primary Examiner(s): Ahmed; Samir A.
• Assistant Examiner(s): Park; Edward
• Application Number:
  US11/299,453
• Publication Number:
  US 20070133874A1
• Time in Patent Office: 1,786 Days
• Field of Search: 382/181, 382/231
• US Class Current: 382/181
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/12 Detection or correction of errors, e.g. by rescanning the pattern
• CPC Class Codes: G06V 30/268 Lexical context

Abstract

In a system for updating a contacts database (42, 46), a portable imager (12) acquires a digital business card image (10). An image segmenter (16) extracts text image segments from the digital business card image. An optical character recognizer (OCR) (26) generates one or more textual content candidates for each text image segment. A scoring processor (36) scores each textual content candidate based on results of database queries respective to the textual content candidates. A content selector (38) selects a textual content candidate for each text image segment based at least on the assigned scores. An interface (50) is configured to update the contacts list based on the selected textual content candidates.

Claims

1. A system for updating a contacts database, the system comprising:

a portable imager configured to acquire a digital business card image;

an image segmenter configured to extract text image segments from the digital business card image;

an optical character recognizer (OCR) configured to generate one or more textual content candidates for each text image segment;

a scoring processor configured to assign scores to two or more textual content candidates generated by the OCR for an ambiguous text image segment based on results of database queries respective to the two or more textual content candidates, wherein the scoring processor comprises;

a local scoring processor component configured to compute a score for each of the two or more textual content candidates based on results of database queries respective to that textual content candidate; and

a global scores adjustment component configured to selectively adjust scores of two or more textual content candidates based on results of database queries respective to textual content candidates generated by the OCR for text image segments other than the ambiguous text image segment;

a content selector that selects one of the two or more textual content candidates based at least on the assigned scores; and

an interface configured to update the contacts database based on the selected one of the two or more textual content candidates.

2. The system as set forth in claim 1 wherein the portable imager includes a cellular telephone including a built-in camera.

3. The system as set forth in claim 1, wherein the portable imager includes a built-in camera of a cellular telephone including a built-in camera.

4. The system as set forth in claim 1, further comprising:

a computer embodying at least the OCR and scoring processor, the computer being physically distinct from the portable imager.

5. The system as set forth in claim 1, wherein at least the image segmenter, OCR, and scoring processor reside on the portable imager.

6. The system as set forth in claim 1, wherein the interface comprises:

a user interface configured to allow an associated user to accept or edit the selected one of the two or more textual content candidates prior to updating the contacts database.

7. A method for acquiring personal information, the method comprising:

acquiring a business card image wherein the acquiring of the business card image comprises photographing the business card using a built in camera of a cellular telephone;

extracting text image segments including an ambiguous text image segment from the business card image;

applying optical character recognition (OCR) to the text image segments including the ambiguous text image segment to generate textual content candidates for the text image segments including two or more textual content candidates for the ambiguous text image segment;

querying at least one database respective to each of the textual content candidates;

assigning scores to the two or more textual content candidates generated by the OCR for the ambiguous text image segment by;

computing a score for each of the two or more textual content candidates generated by the OCR for the ambiguous text image segment based on results of database queries respective to that textual content candidate; and

performing a global adjustment of the scores of the two or more textual content candidates generated by the OCR for the ambiguous text image segment based on results of database queries respective to textual content candidates generated by the OCR for text image segments other than the ambiguous text image segment;

selecting a most likely one of the two or more textual content candidates generated by the OCR for the ambiguous text image segment based at least on the assigned scores;

wherein the extracting, applying, querying, assigning, and selecting are performed by a processor.

8. The method as set forth in claim 7, wherein the computing of scores comprises:

weighting the scores based on a reliability weighting of at least one database from which at least one record is returned.

9. The method as set forth in claim 7, further comprising:

extracting a logo image segment from the business card image; and

querying at least one image database respective to the logo image segment;

the global adjustment being further based on records returned by the querying of the at least one image database respective to the logo image segment.

10. A system for generating a textual contact record from a business card image, the system comprising:

a cellular telephone including a built-in camera configured to (i) acquire an image of a business card using the built-in camera and (ii) extract textual content candidates from the image of the business card by optical character recognition (OCR) and (iii) extract at least one logo image segment from the image of the business card;

a content candidates scoring processor that queries at least one database respective to the textual content candidates and collects records returned responsive to the queries and assigns scores to the textual content candidates based on the collected records, the content candidates scoring processor including a local scoring processor that assigns a score to each textual content candidate based on records returned by queries respective to the textual content candidate;

wherein the content candidates scoring processor also image queries at least one image database respective to the logo image segment and collects textual metadata returned responsive to the image query, the content candidates scoring processor further including a global adjuster that modifies the score of a selected textual content candidate when the collected textual metadata includes the selected textual content candidate; and

a content selector that selects a textual content candidate for each text image segment based at least on the assigned scores.

11. The system as set forth in claim 10, wherein the global adjuster modifies the score of a first textual content candidate that was extracted by OCR from a corresponding first text image segment when at least one record returned by a query respective to a second textual content candidate that was extracted by OCR from a second text image segment different from the first text image segment also includes the first textual content candidate.

12. The system as set forth in claim 10, wherein the at least one database queried by the database query includes a plurality of databases, and the content candidates scoring processor further comprises:

stored database weights corresponding to the databases, the scores assigned based on the collected records incorporating the database weights of the databases from which the records were collected.

13. The system as set forth in claim 12, wherein the content candidates scoring processor further comprises:

a database weights adjuster that adjusts the stored database weights based on ongoing accuracy of support for textual content candidates provided by the databases.

14. The system as set forth in claim 10, wherein the content candidates scoring processor further comprises:

a tagging adjuster that selectively adjusts tags assigned to the text image segments of the business card image based on the collected records.

15. The system as set forth in claim 10, wherein each textual content candidate has a confidence level assigned thereto by the OCR, and the content selector selects the textual content candidate for each text image segment based on the OCR-assigned confidence levels adjusted by the assigned scores.