Scene-based people metering for audience measurement
View Patent ↗Scene-based people metering for audience measurement is disclosed. Example methods disclosed herein include grouping successive image frames depicting a location of a media presentation to form a sequence of scenes, respective scenes including respective groups of the image frames. Such example methods also include grouping matching scenes into respective scene clusters having respective sizes, the scene clusters being represented by respective key frames. Such example methods further include assigning respective ranks to the key frames of the respective scene clusters, the respective ranks being determined based on the respective sizes of the scene clusters. Such example methods additionally include processing the key frames in accordance with the respective ranks to monitor an audience of the media presentation.
1. A method to monitor an audience of a media presentation, the method comprising:
grouping successive image frames depicting a location of the media presentation to form a sequence of scenes, respective scenes including respective groups of the image frames;
grouping matching scenes into respective scene clusters having respective sizes, the scene clusters being represented by respective key frames;
assigning respective ranks to the key frames of the respective scene clusters, the respective ranks being determined based on the respective sizes of the scene clusters; and
processing the key frames in accordance with the respective ranks to monitor the audience of the media presentation.
2. The method as a defined in claim 1 , wherein a first scene cluster includes a first cluster of scenes, and the size of the first scene cluster corresponds to a total number of image frames included among the scenes in the first cluster of scenes.
3. The method as a defined in claim 1 , wherein the key frames respectively representing the scene clusters are first key frames, the scenes are represented by respective second key frames determined from the groups of the image frames included in the respective scenes, and further including determining the first key frames from the second key frames.
4. The method as a defined in claim 3 , wherein a first scene cluster includes a first cluster of scenes, and determining the first key frames from the second key frames includes combining the second key frames that represent the scenes included in the first cluster of scenes to determine the first key frame that is to represent the first scene cluster.
5. The method as a defined in claim 1 , further including obtaining image signatures corresponding to the key frames, wherein grouping the matching scenes includes forming a first scene cluster from a first group of scenes having matching image signatures.
6. The method as a defined in claim 1 , further including assigning a first rank to a first scene cluster that is higher than a second rank assigned to a second scene cluster when the size of the first scene cluster is larger than the size of the second scene cluster.
7. The method as a defined in claim 1 , wherein a first scene cluster includes a first cluster of scenes having at least two scenes that are not adjacent, and processing the key frames includes:
processing a first key frame representing the first scene cluster to determining first audience identification results identifying a first audience depicted in the first key frame; and
associating the first audience identification results with the at least two scenes that are not adjacent.
8. A tangible machine readable medium comprising machine readable instructions that, when executed, cause a machine to at least:
group successive image frames depicting a location of a media presentation to form a sequence of scenes, respective scenes including respective groups of the image frames;
group matching scenes into respective scene clusters having respective sizes, the scene clusters being represented by respective key frames;
assign respective ranks to the key frames of the respective scene clusters, the respective ranks being determined based on the respective sizes of the scene clusters; and
process the key frames in accordance with the respective ranks to monitor an audience of the media presentation.
9. The tangible machine readable medium as defined in claim 8 , wherein a first scene cluster includes a first cluster of scenes, and the size of the first scene cluster corresponds to a total number of image frames included among the scenes in the first cluster of scenes.
10. The tangible machine readable medium as defined in claim 8 , wherein the key frames respectively representing the scene clusters are first key frames, the scenes are represented by respective second key frames determined from the groups of the image frames included in the respective scenes, and the instructions, when executed, further cause the machine to determine the first key frames from the second key frames.
11. The tangible machine readable medium as defined in claim 10 , wherein a first scene cluster includes a first cluster of scenes, and the instructions, when executed, further cause the machine to determine the first key frames from the second key frames by combining the second key frames that represent the scenes included in the first cluster of scenes to determine the first key frame that is to represent the first scene cluster.
12. The tangible machine readable medium as defined in claim 8 , wherein the instructions, when executed, further cause the machine to:
obtain image signatures corresponding to the key frames; and
form a first scene cluster from a first group of scenes having matching image signatures.
13. The tangible machine readable medium as defined in claim 8 , wherein the instructions, when executed, further cause the machine to assign a first rank to a first scene cluster that is higher than a second rank assigned to a second scene cluster when the size of the first scene cluster is larger than the size of the second scene cluster.
14. The tangible machine readable medium as defined in claim 8 , wherein a first scene cluster includes a first cluster of scenes having at least two scenes that are not adjacent, and the instructions, when executed, further cause the machine to:
process a first key frame representing the first scene cluster to determining first audience identification results identifying a first audience depicted in the first key frame; and
associate the first audience identification results with the at least two scenes that are not adjacent.
15. An apparatus to monitor an audience of a media presentation, the apparatus comprising:
a scene change detector to group successive image frames depicting a location of the media presentation to form a sequence of scenes, respective scenes including respective groups of the image frames;
an image clusterer to group matching scenes into respective scene clusters, the scene clusters being represented by respective key frames; and
an audience recognizer to process the key frames to monitor the audience of the media presentation, the audience recognizer to process the key frames in accordance with respective ranks assigned to the key frames, a first rank assigned to a first scene cluster being higher than a second rank assigned to a second scene cluster when a size of the first scene cluster is larger than a size of the second scene cluster.
16. The apparatus as a defined in claim 15 , wherein a first scene cluster includes a first cluster of scenes, and the size of the first scene cluster corresponds to a total number of image frames included among the scenes in the first cluster of scenes.
17. The apparatus as a defined in claim 15 , wherein the key frames representing the scene clusters are first key frames, the scenes are represented by respective second key frames determined from the groups of the image frames included in the respective scenes, and the image clusterer is further to determine the first key frames from the second key frames.
18. The apparatus as a defined in claim 17 , wherein a first scene cluster includes a first cluster of scenes, and the image clusterer is to determine the first key frames from the second key frames by combining the second key frames that represent the scenes included in the first cluster of scenes to determine the first key frame that is to represent the first scene cluster.
19. The apparatus as a defined in claim 15 , wherein the image clusterer is further to:
obtain image signatures corresponding to the key frames; and
form a first scene cluster from a first group of scenes having matching image signatures.
20. The apparatus as a defined in claim 15 , wherein a first scene cluster includes a first cluster of scenes having at least two scenes that are not adjacent, and the audience recognizer is to process the key frames by:
processing a first key frame representing the first scene cluster to determining first audience identification results identifying a first audience depicted in the first key frame; and
associating the first audience identification results with the at least two scenes that are not adjacent.