Method and system for form processing

• US 6,741,724 B1
• Filed: 03/24/2000
• Issued: 05/25/2004
• Est. Priority Date: 03/24/2000
• Status: Expired due to Term

• Current Assignee: Siemens Logistics LLC (Siemens AG)
• Original Assignee: Siemens Dematic Postal Automation LP (Siemens AG)
• Inventors: Kiani, Shahrom, Bruce, Ben F., Kessler, Linda J., Seidel, Gert J., Bishop-Jones, Brenda J.
• Primary Examiner(s): Patel, Jayanti K.
• Assistant Examiner(s): Tabatabai, Abolfazl
• Application Number:
  US09/534,182
• Time in Patent Office: 1,523 Days
• Field of Search: 382/101, 382/112, 382/181, 382/182, 382/183, 382/188, 382/218, 382/219, 382/220, 382/221, 382/309, 382/310, 382/311, 700/219, 705/60, 705/62, 705/401, 705/410, 705/61, 705/407, 705/402, 414/593
• US Class Current: 382/101
• CPC Class Codes: G06F 40/232 Orthographic correction, e.g. spell checking or vowelisation
• CPC Class Codes: G06Q 10/08 Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/127 with the intervention of an operator
• CPC Class Codes: G06V 30/268 Lexical context

Abstract

The invention provides a method of processing an information form such as a postal change of address form. Such a form includes various data elements, such as a name of a person submitting the form and the information on which action is to be taken. In the case of mail forwarding, these data elements would include, for example, the name of the person that wishes mail to be redirected, an old address to which mail for that person was previously sent, and a new address to which mail for that person should be sent. Such a method includes the steps of scanning a sheet having written information thereon reflecting the data elements to obtain an image, processing the image with character recognition logic to obtain text data corresponding to the written information, analyzing the text data with a computer to determine a scanning error probability therein, if the scanning error probability exceeds a predetermined limit, attempting to correct such error by editing of the text data by at least one of a computer using predetermined correction criteria and a human operator viewing the text data, checking one or more elements of the text data (such as a new address and an old address) against a first database of valid data elements (e.g., a postal address database), if the element is found in the first database, updating a second database such as a National Change of Address (NCOA) database to include at least new or revised text data, and rejecting the text data and aborting updating the second database for such text data that cannot be corrected. The invention further provides a system for processing a series of change of information forms such as COA forms, which system is capable of carrying out the foregoing process.

Claims

1. A method of processing an information form having a set of data elements completed by a submitter of the information form, which set of data elements is maintained in a master database, including the steps of:

(a) scanning a sheet having written information thereon reflecting the data elements to obtain an image;

(b) processing the image with character recognition logic to obtain text data corresponding to the written information;

(c) analyzing the text data with a computer to determine a scanning error probability therein;

(d) if the scanning error probability exceeds a predetermined limit, attempting to correct such error by editing of the text data by at least one of a computer using predetermined correction criteria and a human operator viewing the text data;

(e) checking one or more elements of the text data against a verification database containing information derived from the master database to verify that the information form comprises an update to an identified set of data elements in the verification database;

(f) if the information form comprises an update to an identified set of data elements in the verification database updating the master database to include new or revised text data;

(g) rejecting the text data and aborting updating the master database for such text data that cannot be handled in steps (d)-(f).

2. The method of claim 1, wherein step (c) comprises determining an error probability for each character in the text data, and implementing step (d) if the error probability for any one character exceeds the predetermined limit.

3. The method of claim 1, wherein step (c) comprises determining a first error probability for each character in the text data, and implementing step (d) if the first error probability for any one character exceeds the predetermined limit, or if an overall error probability taking into account error probabilities for multiple characters in the text data exceeds a second predetermined limit.

4. The method of claim 1, further comprising:

performing step (b) at least twice using different character recognition logic,

comparing output of each character recognition logic character by character to determine at least two corresponding versions of the text data; and

determining the text data by implementing arbitration rules on the versions of the text data.

5. The method of claim 4, wherein when one version contains a different character than the other version at the same position in the text data, the arbitration rules select the character from the version having a lower error probability than the character from the other version.

6. The method of claim 4, wherein the arbitration rules select the character from the version having a lower error probability than the character from the other version only when:

(a) the character with the lower error probability has an error probability lower than a first, predetermined level, and

(b) the character in the other version at the same position in the text data has an error probability higher than a second, predetermined level which is higher than the first predetermined level.

7. The method of claim 6, wherein the second predetermined level is at least 10% higher than the first predetermined level.

8. The method of claim 1, wherein step (d) comprises attempting to correct such error by editing of the text data by a human operator viewing the text data on a video display and making changes thereto as needed with a text input device.

9. The method of claim 8, wherein the video display further includes at least a portion of the image obtained in step (a).

10. The method of claim 1, wherein step (d) comprises attempting to correct such error by editing of the text data by at least one of a computer using predetermined correction criteria, and if such correction attempt does not result in successful correction according to predefined criteria, then editing the text data by a human operator viewing the text data on a video display and making changes thereto as needed with a text input device.

11. The method of claim 1, wherein the information form is a postal change of address form, the data elements include a name of a person that wishes mail to be redirected, an old address to which mail for that person was previously sent, and a new address to which mail for that person should be sent, the verification database is a database of valid addresses, and the master database is a national change of address database.

12. The method of claim 1, further comprising:

performing step (a) at least twice on each sheet using different scanning conditions; and

performing step (b) at least twice using the output of each scan of step (a);

comparing output of each character recognition logic character by character to determine at least two corresponding versions of the text data, and

determining the text data by implementing arbitration rules on the versions of the text data.

13. The method of claim 1, further comprising a step (h) of notifying the submitter of the form that the form was rejected in step (g).

14. The method of claim 1, wherein the form comprises a handwritten or printed form.

15. The method of claim 1, further comprising:

performing step (b) at least twice using different character recognition logic to obtain two corresponding versions of the text data;

comparing one or more elements of the corresponding versions of the text data to the verification database to determine whether one of the corresponding versions of the text data contains matching elements; and

if one of the corresponding versions includes matching elements, selecting that version for further processing.

16. A method of processing a change of address form which includes data maintained in a master database of valid postal addresses including a name of a person that wishes mail to be redirected, an old address to which mail for that person was previously sent, and a new address to which mail for that person should be sent, comprising:

(a) scanning a sheet having written change of address information thereon to obtain an image;

(b) processing the image with first character recognition logic to obtain text data of the written change of address information;

(c) analyzing the text data with a computer to detect scanning errors therein,

(d) if a scanning error is detected, attempting to correct such error by a human operator viewing the text data on a display and editing the text data;

(e) checking the new address and the old address against a verification database of valid postal addresses derived from the master data base of valid postal addresses to verify that the information form comprises an update to an identified set of data elements in the verification database,

(f) if the new address and the old address are each found in the verification database of valid postal addresses, updating a change of address database to include the text data;

(g) rejecting the text data and aborting updating the change of address database for such text data that cannot be corrected in step (d) or wherein either the new address or the old address cannot be found in the database of valid postal addresses in step (e).

17. The method of claim 16, wherein step (b) further comprises:

processing the image with second character recognition logic different from the first character recognition logic to obtain second text data of the written change of address information;

comparing first text data obtained from the first character recognition logic with the second text data obtained from the second character recognition logic to create a composite text data which is a combination of the first and second text data which includes alternate interpretations of characters which the first and second character recognition logics interpret differently, and

using the composite text data as the text data analyzed in step (c), wherein alternate interpretations of characters comprise scanning errors sent to a human operator in step (d).

18. The method of claim 16, further comprising a step (h) of notifying a submitter of the change of address form that the change of address form was rejected in step (g).

19. The method of claim 18, wherein step (h) further comprises:

sending the notification to the submitter'"'"'s new address if the new address was found in the verification database of valid postal addresses but the old address was not found in the master database of valid postal addresses;

sending the notification to the submitter'"'"'s old address if the old address was found in the verification database of valid postal addresses but the new address was not found in the master database of valid postal addresses.

20. A system for processing a series of information forms having a set of data elements completed by a submitter of the information form, which set of data elements is maintained in a master database, comprising:

a first scanner capable of successively scanning a series of forms having a set of data elements in the form of written information thereon to obtain an image from each form;

first character recognition logic operable on a computer, which character recognition logic generates text data corresponding to the data elements from each image from the first scanner;

error probability logic operable on a computer that analyzes the text data to determine a scanning error probability for the text data;

a verification database containing information derived from the master database to verify that the information form comprises an update to an identified set of data elements in the verification database;

validation logic operable on a computer for checking one or more designated elements of the text data against the verification database, a video coding station usable by a human operator to view and edit the text data, and

handling logic operable on a computer which includes modules for

(i) if the designated data element is found in the verification database, updating a master database to include new or revised text data;

(ii) if the scanning error probability exceeds a certain predetermined limit or the validation logic is unable to find the designated element in the first database, referring the text data to the human operator for review and editing, and

(iii) rejecting the text data and aborting updating the second database for such text data that cannot be corrected by the human operator.

21. The system of claim 20, further comprising:

a first computer which receives a signal containing the scanned image from the scanner and operates the first character recognition logic and the error probability logic, and

a second computer which receives a signal from the first computer containing the text data, which second computer is connected to the video coding station and which operates the handling logic.

22. The system of claim 20, further comprising second character recognition logic operable on a computer, which second character recognition logic is different from the first character recognition logic, which second character recognition logic generates text data corresponding to the written information from each image, and wherein the error probability logic compares output of each character recognition logic character by character to determine at least two corresponding versions of the text data and determines the text data by implementing arbitration rules on the versions of the text data.

23. The system of claim 20, further comprising:

a second scanner capable of successively scanning a series of sheets having written information thereon to obtain an image from each sheet following scanning by the first scanner; and

second character recognition logic operable on a computer, which second character recognition logic generates text data corresponding to the written information from each image from the second scanner,

wherein the error probability logic compares output of each character recognition logic character by character to determine at least two corresponding versions of the text data and determines the text data by implementing arbitration rules on the versions of the text data.

24. A system for processing a series of information forms, comprising:

at least one scanner capable of successively scanning a series of sheets comprising the information forms having written information thereon to obtain an image from each sheet;

first character recognition logic operable on a computer, which first character recognition logic generates text data corresponding to the written information from a scanner image;

second character recognition logic operable on a computer, which second character recognition logic generates text data corresponding to the written information from a scanner image;

error probability logic operable on a computer that analyzes the text data to determine a scanning error probability for the text data, which error probability logic compares output of each character recognition logic character by character to determine at least two corresponding versions of the text data and determines the text data by implementing arbitration rules on the versions of the text data;

a first database of valid data elements;

validation logic operable on a computer for checking one or more designated elements of the text data against the first database;

automatic error detection and correction logic operable on a computer for making corrections to text data that is not validated by the validation logic by comparison of the text data to records of the first database;

a video coding station usable by a human operator to view and edit the text data in the event that the automatic error detection and correction logic fails to successfully correct the text data; and

a second database containing results generated from the automatic error detection and correction logic and the video coding station.

25. The system of claim 24, wherein the form comprises a postal change of address form having data elements for a previous address and a new address.