METHOD AND SYSTEM FOR CLASSIFYING ONE OR MORE IMAGES

• US 20120039539A1
• Filed: 08/18/2011
• Published: 02/16/2012
• Est. Priority Date: 03/08/2010
• Status: Active Grant

• Current Assignee: Vimeo, Inc. (Match Group, Inc.)
• Original Assignee: Magisto Ltd. (Match Group, Inc.)
• Inventors: Rav-Acha, Alex, BOIMAN, OREN
• Granted Patent: US 8,948,515 B2
• Time in Patent Office: Days
• US Class Current: 382/195
• CPC Class Codes: G11B 27/28 by using information signals recorded by the same method as the main recording G11B27/22 takes precedence
• CPC Class Codes: G11B 27/329 on a disc [VTOC]
• CPC Class Codes: G11B 27/34 Indicating arrangements indicating means incorporated in magazine or cassette G11B23/046 and G11B23/0875; indicating measured values in general G01D

Abstract

A method for determining a predictability of a media entity portion, the method includes: receiving or generating (a) reference media descriptors, and (b) probability estimations of descriptor space representatives given the reference media descriptors; wherein the descriptor space representatives are representative of a set of media entities; and calculating a predictability score of the media entity portion based on at least (a) the probability estimations of the descriptor space representatives given the reference media descriptors, and (b) relationships between the media entity portion descriptors and the descriptor space representatives. A method for processing media streams, the method may include: applying probabilistic non-parametric process on the media stream to locate media portions of interest; and generating metadata indicative of the media portions of interest.

Claims

1. A method for classifying at least one image, the method comprises:

partitioning the at least one image to multiple media entities or receiving a partition information indicative of a partition of the at least one image to multiple media entities;

receiving or generating (a) media class descriptors for each media entity class out of a set of media entity classes, and (b) probability estimations of descriptor space representatives given each of the media entity classes;

wherein the descriptor space representatives are representative of a set of media entities;

calculating, for each pair of media entity and media class, a predictability score based on (a) the probability estimations of the descriptor space representatives given the media class descriptors of the media class, and (b) relationships between descriptors of the media entity and the descriptor space representatives;

classifying each media entity of the multiple media entities based on predictability scores of the media entity given each media class; and

providing at least one image classification, based on classes of each of the multiple media entities.

2. The method according to claim 1 wherein the at least one image comprises multiple images and wherein the classifying comprising classifying each image.

3. The method according to claim 1 wherein the at least one image forms a video stream.

4. The method according to claim 1, wherein the at least one image is a single image.

5. The method according to claim 1, comprising applying a human body detection algorithm on the at least one image to provide human body detection results;

and partitioning the at least one image based on the human body detection results.

6. The method according to claim 1, comprising applying a face detection algorithm on the at least one image to provide face detection results;

and partitioning the at least one image based on the face detection results.

7. The method according to claim 1, comprising applying a human skin detection algorithm on the at least one image to provide human skin detection results;

and partitioning the at least one image based on the human skin detection results.

8. The method according to claim 1, comprising determining whether the at least one image should be prevented from being displayed.

9. The method according to claim 1, comprising defining a dominant class and classifying an image out of the at least one image as belonging to the dominant class if at least one media entity of the image is classified as belonging to the dominant class.

10. The method according to claim 1, comprising executing multiple iterations of a sequence of stages that comprises the stages of receiving or generating, calculating, and classifying;

wherein different iterations differ from each other by accuracy of execution and speed of execution.

11. The method according to claim 10, wherein the decision of whether to apply each iteration is based on the output of previous iterations.

12. The method according to claim 10, comprising filtering out an image based on an outcome of at least one iteration of the multiple iterations.

13. The method according to claim 1 comprising:

detecting a human organ;

determining, based on a location of the human organ, an expected location of at least one other human organ; and

verifying the expected location of the at least one other human organ by processing at least one portion of an image that corresponds to the expected location of the at least one other human organ.

14. The method according to claim 1, comprising executing multiple iterations of a sequence of stages that comprises the stages of receiving or generating, calculating, and classifying;

wherein different iterations differ from each other by a complexity level.

15. The method according to claim 14, wherein at least two iterations further comprise partitioning the at least one image.

16. The method according to claim 14, comprising performing multiple iterations, each iteration associated with a higher complexity level.

17. The method according to claim 14, wherein the decision of whether to apply each iteration is based on the output of previous iterations.

18. The method according to claim 1, comprising executing multiple iterations of a sequence of stages that comprises the stages of receiving or generating, calculating, and classifying;

wherein different iterations differ from each other by a number of descriptors of the media entities.

19. The method according to claim 17, wherein at least two iterations further comprise partitioning the at least one image.

20. The method according to claim 10, comprising performing multiple iterations while increasing the number of descriptors of the media entities.

21. The method according to claim 1, comprising executing multiple iterations of a sequence of stages that comprises the stages of receiving or generating, calculating, and classifying;

wherein different iterations differ from each other by a selection of media class descriptors that are taken into account during the iteration.

22. The method according to claim 18, wherein at least two iterations further comprise partitioning the at least one image.

23. The method according to claim 1, comprising executing multiple iterations of a sequence of stages that comprises the stages of receiving or generating, calculating, and classifying;

wherein different iterations differ from each other by descriptors of the media entity that are taken into account during the iteration.

24. The method according to claim 23, wherein at least two iterations further comprise partitioning the at least one image.

25. The method according to claim 1, comprising applying a motion detection algorithm on the at least one image to provide motion detection results;

and partitioning the at least one image based on the motion detection results.

26. A computer program product that comprises a non-transitory computer readable medium that stores instructions for:

partitioning the at least one image to multiple media entities or receiving a partition information indicative of a partition of the at least one image to multiple media entities;

receiving or generating (a) media class descriptors for each media entity class out of a set of media entity classes, and (b) probability estimations of descriptor space representatives given each of the media entity classes;

wherein the descriptor space representatives are representative of a set of media entities;

calculating, for each pair of media entity and media class, a predictability score based on (a) the probability estimations of the descriptor space representatives given the media class descriptors of the media class, and (b) relationships between descriptors of the media entity and the descriptor space representatives;

classifying each media entity of the multiple media entities based on predictability scores of the media entity given each media class; and

providing at least one image classification, based on classes of each of the multiple media entities.