Method and system for analyzing image identifications to identify an entity

• US 8,094,887 B2
• Filed: 03/30/2007
• Issued: 01/10/2012
• Est. Priority Date: 06/08/2006
• Status: Expired due to Fees

• Current Assignee: Kyndryl Incorporated
• Original Assignee: International Business Machines Corporation
• Inventors: Axemo, Sven Georg, Hamilton, Carl Viktor August Johnsson, Sanamrad, Mohammad Ali, Tesdorf, Alexis William Gustaf
• Primary Examiner(s): WERNER, BRIAN P
• Application Number:
  US11/693,944
• Publication Number:
  US 20080031522A1
• Time in Patent Office: 1,747 Days
• Field of Search: 382/105
• US Class Current: 382/105
• CPC Class Codes: G06V 10/987 with the intervention of an operator
• CPC Class Codes: G07B 15/063 using wireless information transmission between the vehicle and a fixed station identification of vehicles G08G
• CPC Class Codes: G08G 1/017 identifying vehicles G08G1/015, G08G1/054 take precedence

Abstract

A method for analyzing image identifications to determine whether image identifications identify an entity (e.g., license plate of a vehicle). Identification sets are received from at least one optical character recognition (OCR) engine. Each identification set includes a character string and an associated confidence level. Each character string is derived by the respective OCR engine from an image of the entity. An identification set is received from each OCR engine. The character strings are compared, resulting in identifying all conflicting character strings, wherein any two non-identical character strings are considered to be conflicting. The confidence level in each identification set is analyzed, resulting in determining whether each confidence level exceeds a predefined threshold confidence level. At least one rule is applied to the preceding results to ascertain whether or not the entity has been identified. An indication of whether or not the entity has been identified is stored.

Claims

1. A method for analyzing image identifications to determine whether the image identifications identify an entity, said method comprising:

receiving, by a validation engine from at least two optical character recognition (OCR) engines, at least two identification sets such that each identification set comprises a character string consisting of a plurality of characters and a confidence level associated with the character string, wherein each character string has been derived by a respective OCR engine of the at least two OCR engines from an image of the entity, wherein at least one identification set of the at least two identification sets is received by the validation engine from each OCR engine;

comparing, by the validation engine, each character string with every other character string, resulting in identifying all conflicting character strings, wherein any two non-identical character strings are considered to be conflicting;

analyzing, by the validation engine, the confidence level in each identification set, resulting in determining whether said each confidence level exceeds a predefined threshold confidence level;

applying, by the validation engine, at least one rule to the identifying resulting from said comparing and/or to the determining resulting from said analyzing, resulting in ascertaining whether or not the entity has been identified; and

storing, in a data store coupled to the validation engine, an indication of whether or not said ascertaining has ascertained that the entity has been identified, wherein said storing is performed by the validation engine.

2. The method of claim 1, wherein said applying the at least one rule results in ascertaining that the entity has not been identified, and wherein the method further comprises:

after said applying the at least one rule, transmitting to a manual agent the at least two identification sets and the images from which the character strings of the at least two identification sets have been derived by the at least two OCR engines, for subsequent use by the manual agent to identify the entity.

3. The method of claim 1, wherein a plurality of identification sets of the at least two identification sets is received by the validation engine from an OCR engine of the at least two OCR engines.

4. The method of claim 1, wherein the at least two identification sets comprises at least four identification sets, wherein a first plurality of identification sets of the at least four identification sets is received by the validation engine from a first OCR engine of the at least two OCR engines, and wherein a second plurality of identification sets of the at least four identification sets is received by the validation engine from a second OCR engine of the at least two OCR engines.

5. The method of claim 1, wherein said comparing determines that at least two character strings are identical and said analyzing determines that the confidence level of each character string of the at least two identical character strings exceeds the predefined threshold confidence level, and wherein said applying the at least one rule results in ascertaining that the entity is identified from the at least two identical character strings.

6. The method of claim 5, wherein the predefined threshold confidence level is 85%.

7. The method of claim 1, wherein said analyzing determines that the confidence level of each character string does not exceed the predefined threshold confidence level, and wherein said applying the at least one rule results in ascertaining that the entity has not been identified.

8. The method of claim 1, wherein said comparing determines that at least two character strings are identical and said analyzing determines that no other character string has a confidence level exceeding the predefined threshold confidence level, and wherein said applying the at least one rule results in ascertaining that the entity is identified from the at least two identical character strings.

9. The method of claim 1, wherein said comparing determines two conflicting character strings and said analyzing determines that the confidence level of each character string of the two conflicting character strings exceeds the predefined threshold confidence level, and wherein said applying the at least one rule results in ascertaining that the entity has not been identified.

10. The method of claim 9, wherein the predefined threshold confidence level is 70%.

11. The method of claim 1, wherein said analyzing determines that the confidence level of each character string of at least two character strings exceeds the predefined threshold confidence level and said comparing determines that two character strings of the at least two character strings are non-identical, and wherein said applying the at least one rule results in ascertaining that the entity has not been identified.

12. The method of claim 1, wherein said analyzing determines that the confidence level of only one character string exceeds the predefined threshold confidence level, and wherein said applying the at least one rule results in ascertaining that the entity has not been identified.

13. The method of claim 12, wherein the predefined threshold confidence level is 85%.

14. A data processing system comprising a processor and a computer readable memory device coupled to the processor, said memory device containing software code configured to be executed by the processor to perform the method of claim 1, wherein the data processing system further comprises the at least two OCR engines, the verification engine, and the data store.

15. A computer program product, comprising a computer readable storage device having software code that when executed on a computer of a data processing system implements the method of claim 1.

16. A method for analyzing image identifications to determine whether the image identifications identify a license plate of a vehicle, said method comprising:

receiving, by a validation engine from at least one optical character recognition (OCR) engine, at least two identification sets such that each identification set comprises a character string consisting of a plurality of characters and a confidence level associated with the character string, wherein each character string has been derived by a respective OCR engine of the at least one OCR engine from an image of the license plate attached to the front or rear of the vehicle, wherein at least one identification set of the at least two identification sets is received by the validation engine from each OCR engine;

comparing, by the validation engine, each character string with every other character string, resulting in identifying all conflicting character strings, wherein any two non-identical character strings are considered to be conflicting;

analyzing, by the validation engine, the confidence level in each identification set, resulting in determining whether said each confidence level exceeds a predefined threshold confidence level and/or is within a predefined range of confidence levels;

applying, by the validation engine, at least one rule to the identifying resulting from said comparing and/or to the determining resulting from said analyzing, resulting in ascertaining whether or not the license plate has been identified; and

storing, in a data store coupled to the validation engine, an indication of whether or not said ascertaining has ascertained that the license plate has been identified, wherein said storing is performed by the validation engine.

17. The method of claim 16,

wherein the at least one OCR engine consists of a single OCR engine;

wherein the character string of a first identification set of the at least two identification sets has been derived by the single OCR engine from an image of the license plate attached to the front of the vehicle, and

wherein the character string of a second identification set of the at least two identification sets has been derived by the single OCR engine from an image of the license plate attached to the rear of the vehicle.

18. The method of claim 16, wherein the at least one OCR engine consists of at least two OCR engines.

19. The method of claim 18, wherein each character string has been derived by the respective OCR engine from an image of the license plate attached to the front of the vehicle or each character string has been derived by the respective OCR engine from an image of the license plate attached to the rear of the vehicle.

20. The method of claim 18, wherein said applying the at least one rule results in ascertaining that the license plate has not been identified, and wherein the method further comprises:

after said applying the at least one rule, transmitting to a manual agent the at least two identification sets and the images from which the character strings of the at least two identification sets have been derived by the at least two OCR engines, for subsequent use by the manual agent to identify the license plate.

21. The method of claim 18, wherein a plurality of identification sets of the at least two identification sets is received by the validation engine from an OCR engine of the at least two OCR engines.

22. The method of claim 18, wherein the at least two identification sets comprises at least four identification sets, wherein a first plurality of identification sets of the at least four identification sets is received by the validation engine from a first OCR engine of the at least two OCR engines, and wherein a second plurality of identification sets of the at least four identification sets is received by the validation engine from a second OCR engine of the at least two OCR engines.

23. The method of claim 18, wherein said comparing determines that at least two character strings are identical and said analyzing determines that the confidence level of each character string of the at least two identical character strings exceeds the predefined threshold confidence level, and wherein said applying the at least one rule results in ascertaining that the license plate is identified from the at least two identical character strings.

24. The method of claim 18, wherein said analyzing determines that the confidence level of each character string does not exceed the predefined threshold confidence level, and wherein said applying the at least one rule results in ascertaining that the license plate has not been identified.

25. The method of claim 18, wherein said comparing determines that at least two character strings are identical and said analyzing determines that no other character string has a confidence level exceeding the predefined threshold confidence level, and wherein said applying the at least one rule results in ascertaining that the license plate is identified from the at least two identical character strings.

26. The method of claim 25, wherein the predefined threshold confidence level is 70%.

27. The method of claim 18, wherein said comparing determines two conflicting character strings and said analyzing determines that the confidence level of each character string of the two conflicting character strings exceeds the predefined threshold confidence level, and wherein said applying the at least one rule results in ascertaining that the license plate has not been identified.

28. The method of claim 27, wherein the predefined threshold confidence level is 70%.

29. The method of claim 18, wherein said analyzing determines that the confidence level of each character string of at least two character strings exceeds the predefined threshold confidence level and said comparing determines that two character strings of the at least two character strings are non-identical, and wherein said applying the at least one rule results in ascertaining that the license plate has not been identified.

30. The method of claim 18, wherein said analyzing determines that the confidence level of only one character string exceeds the predefined threshold confidence level, and wherein said applying the at least one rule results in ascertaining that the license plate has not been identified.

31. The method of claim 18, wherein the at least two identification sets comprises three identification sets, wherein said analyzing determines that the confidence level of each character string of the three identification sets exceeds the predefined threshold confidence level and does not exceed a predefined maximum confidence level and said comparing determines that the character strings of the three identification sets are identical and the character strings of the three identification sets have been derived by the respective OCR engine from images of the license plate attached to both the front and rear of the vehicle, and wherein said applying the at least one rule results in ascertaining that the license plate is identified from the identical character strings of the three identification sets.

32. The method of claim 18, wherein the predefined threshold confidence level is 70%, and wherein the predefined maximum confidence level is 85%.

33. The method of claim 18, wherein said analyzing determines that the confidence level of one character string exceeds the predefined threshold confidence level and the one character string identifies the vehicle as a foreign vehicle, and wherein said applying the at least one rule results in ascertaining that the license plate has not been identified.

34. A data processing system comprising a processor and a computer readable memory device coupled to the processor, said memory device containing software code configured to be executed by the processor to perform the method of claim 16, wherein the data processing system further comprises the at least two OCR engines, the verification engine, and the data store.

35. A computer program product, comprising a computer readable storage device having software code that when executed on a computer of a data processing system implements the method of claim 16.