METHODS AND APPARATUS TO MEASURE BRAND EXPOSURE IN MEDIA STREAMS

• US 20090123025A1
• Filed: 09/26/2008
• Published: 05/14/2009
• Est. Priority Date: 11/09/2007
• Status: Active Grant

• Current Assignee: Nielsen Company (US), LLC
• Original Assignee: Nielsen Company (US), LLC
• Inventors: Deng, Kevin Keqiang, Greenberg, Barry
• Granted Patent: US 9,239,958 B2
• Time in Patent Office: Days
• US Class Current: 382/100
• CPC Class Codes: G06V 20/40 in video content extracting overlay text G06V20/62; video retrieval G06F16/70; processing of video elementary streams in video servers H04N21/234; processing of video elementary streams in video clients H04N21/44
• CPC Class Codes: G06V 20/44 Event detection
• CPC Class Codes: G06V 20/635 Overlay text, e.g. embedded captions in a TV program
• CPC Class Codes: G06V 2201/09 Recognition of logos

Abstract

Methods and apparatus to measure brand exposure in media streams are disclosed. An example method to determine brand exposures included in media content disclosed herein comprises determining whether a scene detected from a media stream corresponding to the media content matches a reference scene, identifying an expected region of interest in the detected scene based on information describing the reference scene when the detected scene is determined to match the reference scene and the reference scene is not specified to be a scene of no interest, and determining whether a reference brand identifier associated with the reference scene is included in the expected region of interest identified in the detected scene.

Claims

1. A method to determine brand exposures included in media content, the method comprising:

determining whether a scene detected from a media stream corresponding to the media content matches a reference scene;

identifying an expected region of interest in the detected scene based on information describing the reference scene when;

(a) the detected scene is determined to match the reference scene and (b) the reference scene is not specified to be a scene of no interest; and

determining whether a reference brand identifier associated with the reference scene is included in the expected region of interest identified in the detected scene.

6. A method as defined in claim 1 further comprising identifying a sequence of adjacent video image frames having substantially similar characteristics and wherein the detected scene comprises the identified sequence of adjacent video image frames having substantially similar characteristics.

9. A method as defined in claim 6 wherein identifying the sequence of adjacent video image frames comprises comparing signatures representative of the video image frames to a key signature representative of the detected scene to determine when a scene change occurs in the media stream.

10. A method as defined in claim 9 wherein the key signature is a first key signature and further comprising determining that a particular video image frame in the sequence of video image frames corresponds to a starting video image frame in the sequence of video image frames comprising the detected scene when a signature corresponding to the particular video image frame differs from a second key signature representative of an immediately preceding detected scene based on a threshold.

11. A method as defined in claim 9 further comprising determining that a particular video image frame in the sequence of video image frames corresponds to an ending video image frame in the sequence of video image frames comprising the detected scene when a signature corresponding to an immediately subsequent video image frame relative to the particular video image frame differs from the key signature based on a threshold.

13. A method as defined in claim 9 wherein the key signature comprises a statistical combination of a plurality of signatures corresponding to a plurality of video image frames in the sequence of video image frames.

14. A method as defined in claim 9 wherein each of the plurality of signatures corresponds to a histogram of at least one of luminance values or chrominance values associated with a respective video image frame.

15. A method as defined in claim 6 further comprising determining a key frame representative of the identified sequence of video image frames comprising the detected scene.

17. A method as defined in claim 15 wherein the key frame comprises a statistical combination of a plurality of video image frames in the sequence of video image frames.

18. A method as defined in claim 1 wherein determining whether the detected scene matches the reference scene comprises comparing a key signature representative of the detected scene with a reference signature representative of the reference scene, wherein the comparison is based on a threshold.

19. A method as defined in claim 18 wherein the threshold is a first threshold different from a second threshold used to compare signatures corresponding to a sequence of video image frames to determine whether at least a portion of the sequence of video image frames constitute the detected scene.

28. A method as defined in claim 1 further comprising prompting a user to specify a brand identifier included in the detected scene when at least one of the reference brand identifier associated with the reference scene is determined to not be included in the expected region of interest or the detected scene is determined to not match the reference scene, and enabling the user to specify a new region of interest including at least one brand identifier in a key video image frame representative of the detected scene.

29. A method as defined in claim 28 further comprising determining the detected scene corresponds to a scene of no interest when no new regions of interest are identified by the user.

30. A method as defined in claim 28 further comprising tracking at least one new region of interest specified by the user across video image frames comprising the detected scene to determine whether an occlusion at least one of changes or lessens and, if the occlusion at least one of changes or lessens, determining at least one of a composite region of interest or a composite reference brand identifier template comprising at least portions of the at least one new region of interest and at least portions of the video image frames in which the occlusion was determined to have at least one of changed or lessened.

34. A method as defined in claim 1 wherein the reference scene is a first reference scene and the information is first information, and further comprising storing second information representative of the detected scene to allow the detected scene to be a second reference scene in a subsequent scene matching attempt.

37. A method as defined in claim 1 further comprising tracking the expected region of interest across video image frames comprising the detected scene when the reference brand identifier is determined to be included in the expected region of interest.

38. A method as defined in claim 37 wherein tracking the expected region of interest comprises comparing a neighborhood including the expected region of interest in a key frame representative of the detected scene with similar neighborhoods in at least some of the video image frames comprising the detected scene to determine at least one appearance parameter characterizing a display of the reference brand identifier across the video image frames comprising the detected scene.

39. A method as defined in claim 38 wherein the at least one appearance parameter comprises at least one of a location, a size, a matching quality or a visual quality of the reference brand identifier.

40. A method as defined in claim 37 further comprising discontinuing tracking after at least one of an ending video frame of the detected scene is compared with the key frame or the reference brand identifier is not detected in a number of the image frames comprising the detected scene.

2. (canceled)

3. (canceled)

4. (canceled)

5. (canceled)

7. (canceled)

8. (canceled)

12. (canceled)

16. (canceled)

20. (canceled)

21. (canceled)

22. (canceled)

23. (canceled)

24. (canceled)

25. (canceled)

26. (canceled)

27. (canceled)

31. (canceled)

32. (canceled)

33. (canceled)

35. (canceled)

36. (canceled)

41. (canceled)

42. (canceled)

43. (canceled)

44. (canceled)

45. (canceled)

46. (canceled)

47. (canceled)

48. (canceled)

49. (canceled)

50. (canceled)

51. (canceled)

52. An apparatus to measure brand exposure associated with media content provided by a media stream, the apparatus comprising:

a scene recognizer comprising;

a scene detector to divide the media stream into a plurality of detected scenes, each detected scene comprising a sequence of image frames having substantially similar signatures; and

a scene classifier to compare a detected scene in the plurality of detected scenes to a plurality of reference scenes to determine whether to classify the detected scene as at least one of a scene of interest or a scene of no interest based on stored information associated with a matching reference scene; and

a brand recognizer comprising;

a region of interest detector to identify an expected region of interest in the detected scene from the stored information associated with the matching reference scene when the scene recognizer classifies the detected scene as a scene of interest; and

a brand identifier matcher to compare the expected region of interest identified in the detected scene to a plurality of templates corresponding to a reference brand identifier to determine whether the reference brand identifier is presented in the expected region of interest.

53. An apparatus as defined in claim 52 further comprising a signature generator to determine signatures representative of image frames corresponding to the media stream, the signatures comprising histograms of at least one of luminance or chrominance values included in the image frames, and the scene detector being configured to determine whether first and second image frames have substantially similar signatures by

computing differences between histogram values of a first signature representative of the first image frame and respective histogram values of a second signature representative of the second image frame;

comparing a sum of the computed differences to a first threshold; and

determining that the first and second image frames have substantially similar signatures when the sum is within the first threshold.

54. An apparatus as defined in claim 53 wherein the scene detector is further configured to determine a key signature representative of the detected scene, and wherein the scene classifier is further configured to:

determine a plurality of differences between the key signature and a plurality of reference signatures representative of the plurality of reference scenes; and

determine whether the detected scene matches at least one of the reference scenes based on comparing the plurality of determined differences to a second threshold different from the first threshold.

55. An apparatus as defined in claim 52 wherein scene classifier is further configured to classify the detected scene to be at least one of:

a scene of no interest when the detected scene is determined to match a first reference scene previously classified to be a scene of no interest;

a scene of interest when the detected scene is determined to match a second reference scene previously classified to be a scene of interest;

or

a new scene when the detected scene fails to match any of the plurality of reference scenes.

63. An apparatus as defined in claim 52 wherein the brand recognizer further comprises a tracking module to track the expected region of interest across at least some of the sequence of image frames comprising the detected scene when the reference brand identifier is determined to be included in the expected region of interest.

64. An apparatus as defined in claim 63 wherein the tracking module is configured to compare a neighborhood including the expected region of interest in a key frame representative of the detected scene with similar neighborhoods in the at least some of the sequence of image frames comprising the detected scene to determine at least one appearance parameter characterizing a display of the reference brand identifier across the sequence of image frames comprising the detected scene.

65. An apparatus as defined in claim 64 wherein the at least one appearance parameter comprises at least one of a location, a size, a matching quality or a visual quality of the reference brand identifier.

56. (canceled)

57. (canceled)

58. (canceled)

59. (canceled)

60. (canceled)

61. (canceled)

62. (canceled)

66. (canceled)

67. A system to measure brand exposure associated with media content, the system comprising:

a brand exposure monitor to determine brand identifier information representative of brand identifiers embedded in expected regions of interest in detected scenes of the media content, an expected region of interest being identified in a detected scene based on comparing the detected scene with a plurality of reference scenes to classify the detected scene; and

an audience measurement system to determine audience exposure information measuring audience exposure of the media content and to combine the embedded brand identifier information and the audience exposure information to determine brand exposure information measuring audience exposure to the brand identifiers.

68. A system as defined in claim 67 wherein the exposure exposure information comprises a statistical value representative of a number of audience members exposed to at least one brand identifier.

69. A system as defined in claim 67 wherein the brand exposure monitor automatically:

determines a detected signature for the detected scene;

compares the detected signature to a plurality of reference signatures corresponding respectively to the plurality of reference scenes;

when the detected signature matches a reference signature, classifies the detected scene according to a predetermined classification of a matching reference scene corresponding to the matching reference signature;

when the brand exposure monitor classifies the detected scene to be a scene of interest, identifies the expected region of interest in the detected scene based on a reference region of interest associated with the matching reference scene; and

compares the expected region of interest in the detected scene to a plurality of templates corresponding to a reference brand identifier associated with a reference region of interest in the matching reference scene to determine whether the reference brand identifier is embedded in the detected scene.

70. (canceled)