Scene-based people metering for audience measurement

1. A method to perform people metering for audience measurement, the method comprising:

segmenting image frames depicting a location in which an audience is expected to be present to form a sequence of scenes;

for a first scene in the sequence of scenes, identifying a first key frame representing a first sequence of image frames corresponding to the first scene; and

processing the first key frame to identify an audience depicted in the first scene, wherein processing the first key frame comprises:

comparing an image signature representative of the first key frame with a set of reference signatures representative of a set of reference scenes to determine a comparison result; and

processing the first key frame based on the comparison result to identify the audience depicted in the first scene, wherein processing the first key frame based on the comparison result comprises:

assigning the first key frame to a processing queue with a first priority when the comparison result indicates that the image signature does not match at least one reference signature in the set of reference signatures;

assigning the first key frame to the processing queue with a second priority when the comparison result indicates that the image signature matches at least one reference signature in the set of reference signatures, the second priority being lower than the first priority; and

using priorities associated with key frames included in the processing queue to prioritize performing processing on the key frames to identify respective audiences depicted in respective scenes represented by the key frames, the key frames including the first key frame.

2. A method as defined in claim 1 , wherein segmenting the sequence of image frames comprises:

determining image signatures representative of the image frames; and

grouping successive frames having similar image signatures into respective sequences of the image frames, respective sequences of image frames forming respective scenes in the sequence of scenes.

3. A method as defined in claim 1 , wherein the first key frame comprises at least one of a starting image frame, a midpoint image frame, an ending image frame or a statistical combination of the first sequence of image frames forming the first scene.

4. A method as defined in claim 1 , wherein processing the first key frame further comprises performing facial recognition processing on the first key frame to identify the audience depicted in the first scene.

5. A method as defined in claim 1 , wherein processing the first key frame based on the comparison result further comprises, when the comparison result indicates that the image signature matches a first reference signature in the set of reference signatures, using audience identification information associated with a first reference scene represented by the first reference signature to identify the audience depicted in the first scene.

6. A method as defined in claim 1 , further comprising:

associating audience identification information with the image signature; and

including the image signature in the set of reference signatures for comparison with a second image signature representative of a subsequent scene occurring after the first scene.

7. A tangible machine readable medium comprising machine readable instructions which, when executed, cause a machine to at least:

segment image frames depicting a location in which an audience is expected to be present to form a sequence of scenes;

for a first scene in the sequence of scenes, identify a first key frame representing a first sequence of image frames corresponding to the first scene; and

process the first key frame to identify an audience depicted in the first scene, wherein to process the first key frame, the machine readable instructions, when executed, further cause the machine to:

compare an image signature representative of the first key frame with a set of reference signatures representative of a set of reference scenes to determine a comparison result; and

process the first key frame based on the comparison result to identify the audience depicted in the first scene, wherein the machine readable instructions, when executed, further cause the machine to:

assign the first key frame to a processing queue with a first priority when the comparison result indicates that the image signature does not match at least one reference signature in the set of reference signatures;

assign the first key frame to the processing queue with a second priority when the comparison result indicates that the image signature matches at least one reference signature in the set of reference signatures, the second priority being lower than the first priority; and

use priorities associated with key frames included in the processing queue to prioritize performing processing on the key frames to identify respective audiences depicted in respective scenes represented by the key frames, the key frames including the first key frame.

8. A tangible machine readable medium as defined in claim 7 , wherein the first key frame comprises at least one of a starting image frame, a midpoint image frame, an ending image frame or a statistical combination of the first sequence of image frames forming the first scene.

9. A tangible machine readable medium as defined in claim 7 , wherein to process the first key frame, the machine readable instructions, when executed, further cause the machine to perform facial recognition processing on the first key frame to identify the audience depicted in the first scene.

10. A tangible machine readable medium as defined in claim 7 , wherein to segment the sequence of image frames, the machine readable instructions, when executed, further cause the machine to:

determine image signatures representative of the image frames; and

group successive frames having similar image signatures into respective sequences of image frames, respective sequences of image frames forming respective scenes in the sequence of scenes.

11. A tangible machine readable medium as defined in claim 7 , wherein the machine readable instructions, when executed, further cause the machine to use audience identification information associated with a first reference scene represented by a first reference signature to identify the audience depicted in the first scene when the comparison result indicates that the image signature matches the first reference signature.

12. A tangible machine readable medium as defined in claim 7 , wherein the machine readable instructions, when executed, further cause the machine to:

associate audience identification information with the image signature; and

include the image signature in the set of reference signatures for comparison with a second image signature representative of a subsequent scene occurring after the first scene.

13. An apparatus to identify people in an audience, the apparatus comprising:

a scene change detector to segment image frames depicting a location in which an audience is expected to be present to form a sequence of scenes; and

an audience recognizer to process a first key frame representing a first sequence of image frames corresponding to a first scene in the sequence of scenes to identify an audience depicted in the first scene, wherein the audience recognizer is further to:

compare an image signature representative of the first key frame with a set of reference signatures representative of a set of reference scenes to determine a comparison result; and

process the first key frame based on the comparison result to identify the audience depicted in the first scene;

assign the first key frame to a processing queue with a first priority when the comparison result indicates that the image signature does not match at least one reference signature in the set of reference signatures;

assign the first key frame to the processing queue with a second priority when the comparison results indicates that the image signature matches at least one reference signature in the set of reference signatures, the second priority being lower than the first priority; and

use priorities associated with key frames included in the processing queue to prioritize performing processing on the key frames to identify respective audiences depicted in respective scenes represented by the key frames, the key frames including the first key frame.

14. An apparatus as defined in claim 13 , wherein the scene change detector is to:

determine image signatures representative of the image frames; and

include successive frames having similar image signatures into respective sequences of image frames, respective sequences forming respective scenes in the sequence of scenes.

15. An apparatus as defined in claim 13 , wherein the first key frame comprises at least one of a starting image frame, a midpoint image frame, an ending image frame or a statistical combination of the first sequence of image frames forming the first scene.

16. An apparatus as defined in claim 13 , wherein the audience recognizer is to use first audience identification information associated with a first reference scene represented by a first reference signature to identify the audience depicted in the first scene when the comparison result indicates that the image signature matches the first reference signature.

17. An apparatus as defined in claim 13 , wherein to process the first key frame, the audience recognizer is further to perform facial recognition processing on the first key frame to identify the audience depicted in the first scene.

18. An apparatus as defined in claim 13 , wherein the audience recognizer is further to:

associate audience identification information with the image signature; and

include the image signature in the set of reference signatures for comparison with a second image signature representative of a subsequent scene occurring after the first scene.