Granted Patent B2

US 11,080,318 · App. 14/310,777 · Granted Aug 3, 2021

Inventors: Madirakshi Das (Penfield, NY); Alexander C. Loui (Penfield, NY)

Assignee: KODAK ALARIS INC.

G06F16/41

Abstract

Claims (40)

1. A method for ranking events in media collections, performed at a device having one or more processors and a processor-accessible memory, the method comprising:

designating a media collection stored on the processor-accessible memory, wherein the media collection comprises a plurality of media collection items, and wherein the media collection items capture a plurality of events;

clustering the media collection items into a hierarchical event structure, wherein the hierarchical event structure comprises events and sub-events;

identifying visually similar sub-events within each event in the hierarchical event structure, wherein the visually similar sub-events have consistent color distribution and are identified by block-level color histogram similarity;

ranking the events in descending order of importance based on a count of most to fewest visually similar sub-events within each event;

identifying a plurality of event attributes present in the ranked events;

receiving a selection from a user of a preferred event attribute;

determining a threshold target distribution of the preferred event attribute, wherein determining the threshold target distribution comprises generating a normalized histogram having a plurality of bins, wherein each bin represents a category of the preferred event attribute;

selecting a subset of ranked events that combined meet or exceed the threshold target distribution of the preferred event attribute;

selecting a first event from the subset of ranked events with a highest ranking of importance;

selecting media collection items that capture the selected first event, wherein the number of media collection items selected is based on the threshold target distribution of the preferred event attribute;

selecting a second event from the subset of ranked events;

selecting media collection items that capture the selected second event, wherein the number of media collection items selected is based on the threshold target distribution of the preferred event attribute; and

fulfilling an output product with the selected media collection items.

2. The method of claim 1 , wherein ranking the events is also based on a significance score of the event.

3. The method of claim 1 , wherein ranking the events is also based on a distribution that models importance of an event over an elapsed time period.

4. The method of claim 1 , wherein ranking the events is also based on a score or distribution that models interestingness of an event over an elapsed period of time.

5. The method of claim 1 , wherein ranking the events is also based on metadata from social networks.

6. The method of claim 1 , wherein ranking the events is also based on metadata from social networks through analyzing user tags and comments.

7. The method of claim 1 , wherein ranking the events is also based on the media collection items capturing the events that have been marked by a user as being a favorite or to be used for sharing.

8. A method for selecting events from media collections, comprising:

designating a media collection, wherein the media collection comprises a plurality of images;

using a processor to cluster the media collection items into a hierarchical event structure, wherein the hierarchical event structure comprises events and sub-events;

using the processor to identify and count visually similar sub-events within each event in the hierarchical event structure, wherein the visually similar sub-events have consistent color distribution and are identified by block-level color histogram similarity;

using the processor to determine a ranked list of the events in the hierarchical event structure, wherein the ranked list prioritizes the events from highest interestingness score to lowest interestingness score, wherein interestingness score is determined based on time elapsed since each of the events in the hierarchical event structure;

receiving a selection from a user of a preferred event attribute;

using the processor to calculate a threshold target distribution of the preferred event attribute, wherein calculating the threshold target distribution comprises generating a normalized histogram having a plurality of bins, wherein each bin represents a category of the preferred event attribute;

selecting a subset of ranked events that combined meet or exceed the threshold target distribution of the preferred event attribute;

selecting a first event from the subset of ranked events with a highest interestingness score ranking;

selecting images from the media collection that capture the selected first event, wherein the number of images selected is based on the threshold target distribution of the preferred event attribute;

selecting a second event from the subset of ranked events;

selecting images from the media collection that capture the selected second event, wherein the number of images selected is based on the threshold target distribution of the preferred event attribute; and

using the processor to incorporate the selected images from the media collection into an output product.

9. The method of claim 8 , wherein the preferred event attribute is event class.

10. The method of claim 8 , wherein the preferred event attribute is event size.

11. The method of claim 8 , wherein the preferred event attribute is media type of the event.

12. The method of claim 8 , wherein determining the ranked list of events is based on a significance score of each event.

13. The method of claim 8 , wherein determining the ranked list of events is based on metadata from social networks.

14. The method of claim 8 , wherein determining the ranked list of events is based on metadata from social networks through analyzing user tags and comments.

15. The method of claim 8 , wherein determining the ranked list of events is based on a number of images capturing the event that have been marked by a user as being a favorite or to be used for sharing.

Assignments (9)

Continuity (2)