EVENT DETECTION APPARATUS AND EVENT DETECTION METHOD

• US 20130121590A1
• Filed: 11/08/2012
• Published: 05/16/2013
• Est. Priority Date: 11/10/2011
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Yamanaka, Masao, Matsugu, Masakazu, Katano, Yasuo
• Granted Patent: US 9,824,296 B2
• Time in Patent Office: Days
• US Class Current: 382/195
• CPC Class Codes: G06F 18/22 Matching criteria, e.g. proximity measures
• CPC Class Codes: G06V 10/758 Involving statistics of pixels or of feature values, e.g. histogram matching
• CPC Class Codes: G06V 20/47 Detecting features for summarising video content
• CPC Class Codes: G06V 20/52 Surveillance or monitoring of activities, e.g. for recognising suspicious objects recognising microscopic objects G06V20/69
• CPC Class Codes: H04N 5/144 Movement detection for video coding H04N19/503; analysis of motion in general G06T7/20
• CPC Class Codes: H04N 5/147 Scene change detection

Abstract

An event detection apparatus includes an input unit configured to input a plurality of time-sequential images, a first extraction unit configured to extract sets of first image samples according to respective different sample scales from a first time range of the plurality of time-sequential images based on a first scale parameter, a second extraction unit configured to extract sets of second image samples according to respective different sample scales from a second time range of the plurality of time-sequential images based on a second scale parameter, a dissimilarity calculation unit configured to calculate a dissimilarity between the first and second image samples based on the sets of the first and second image samples, and a detection unit configured to detect an event from the plurality of time-sequential images based on the dissimilarity.

Claims

1. An event detection apparatus comprising:

an input unit configured to input a plurality of time-sequential images;

a first extraction unit configured to extract sets of first image samples according to respective different sample scales from a first time range of the plurality of time-sequential images based on a first scale parameter;

a second extraction unit configured to extract sets of second image samples according to respective different sample scales from a second time range of the plurality of time-sequential images based on a second scale parameter;

a dissimilarity calculation unit configured to calculate a dissimilarity between the first and second image samples based on the sets of the first and second image samples; and

a detection unit configured to detect an event from the plurality of time-sequential images based on the dissimilarity.

2. The event detection apparatus according to claim 1, wherein the first extraction unit is configured to extract the sets of the first image samples from images generated before a predetermined image among the plurality of time-sequential images, and

wherein the second extraction unit is configured to extract the sets of the second image samples from images generated after the predetermined image among the plurality of time-sequential images.

3. The event detection apparatus according to claim 1, wherein the first and second scale parameters include the first and second time ranges, respectively.

4. The event detection apparatus according to claim 1, wherein the first and second scale parameters include the number of image samples in the set of the first image samples and the number of image samples in the set of the second image samples, respectively.

5. The event detection apparatus according to claim 1, wherein the first and second scale parameters include a sampling interval in the set of the first image samples and a sampling interval in the set of the second image samples, respectively.

6. The event detection apparatus according to claim 1, wherein the first and second scale parameters include the number of the sets of the first image samples and the number of the sets of the second image samples, respectively.

7. The event detection apparatus according to claim 1, further comprising:

a setting unit configured to set the first and second scale parameters.

8. The event detection apparatus according to claim 1, wherein the detection unit is configured to detect an event, based on a plurality of dissimilarity patterns acquired from a predetermined reference image, from a pattern of the dissimilarity between respective corresponding first and second image samples in the sets of the first and second image samples.

9. The event detection apparatus according to claim 1, wherein the dissimilarity calculation unit is configured to calculate the dissimilarity based on a ratio between probability densities of predetermined feature quantities respectively calculated from corresponding first image samples and second image samples in the sets of the first image samples and the sets of the second image samples.

10. The event detection apparatus according to claim 1, wherein the detection unit includes a likelihood calculation unit configured to calculate a likelihood of occurrence of an event in a specific image among the plurality of time-sequential images based on the dissimilarity, and a determination unit configured to determine whether the event occurs in the specific image based on the likelihood.

11. The event detection apparatus according to claim 10, wherein the determination unit is configured to determine that the event occurs in the specific image if the likelihood is higher than a predetermined threshold value.

12. The event detection apparatus according to claim 10, wherein the likelihood calculation unit is configured to calculate the likelihood for a plurality of specific images, and

wherein the determination unit is configured to determine that the event occurs in the specific image having a maximum likelihood.

13. An event detection apparatus comprising:

an input unit configured to input a plurality of time-sequential images;

a first extraction unit configured to extract sets of first image samples according to respective different sample scales from the plurality of time-sequential images based on a scale parameter;

a similarity calculation unit configured to calculate, based on the sets of the first image samples and sets of second image samples provided in advance, a similarity between the first image samples and the second image samples; and

a detection unit configured to detect an event from the plurality of time-sequential images based on the similarity.

14. The event detection apparatus according to claim 13, wherein the sets of the second image samples include an image in which a predetermined event occurs, and

wherein the detection unit is configured to detect the predetermined event from the plurality of time-sequential images.

15. An event detection method comprising:

inputting a plurality of time-sequential images;

extracting sets of first image samples according to respective different sample scales from a first time range of the plurality of time-sequential images based on a first scale parameter;

extracting sets of second image samples according to respective different scales from a second time range of the plurality of time-sequential images based on a second scale parameter;

calculating a dissimilarity between the first and second image samples based on the sets of the first and second image samples; and

detecting an event from the plurality of time-sequential images based on the dissimilarity.

16. A storage medium storing a program that causes a computer to execute the event detection method according to claim 15.

17. An event detection method comprising:

inputting a plurality of time-sequential images;

extracting sets of first image samples according to respective different sample scales from the plurality of time-sequential images based on a scale parameter;

calculating, based on the sets of the first image samples and sets of second image samples provided in advance, a similarity between the first image samples and the second image samples; and

detecting an event from the plurality of time-sequential images based on the similarity.

18. A storage medium storing a program that causes a computer to execute the event detection method according to claim 17.

19. An action recognition apparatus configured to recognize an action of an object in a plurality of time-sequential images, the action recognition apparatus comprising:

a first image sample extraction unit configured to extract a first image sample from a plurality of time-sequential images belonging to a known action category;

a second image sample extraction unit configured to extract a second image sample from a plurality of time-sequential images that is a recognition target;

a feature quantity calculation unit configured to calculate feature quantities for the first image sample and feature quantities for the second image sample;

a connection unit configured to generate a first connected feature quantity and a second connected feature quantity by temporally connecting the feature quantities for the first image sample and the feature quantities for the second image sample, respectively;

a similarity calculation unit configured to calculate a similarity between the first connected feature quantity and the second connected feature quantity; and

a recognition unit configured to recognize the action of the object based on the similarity.

20. The action recognition apparatus according to claim 19, wherein the connection unit is configured to generate a plurality of connected feature quantities in such a manner that consecutive feature quantities partially overlap while maintaining a temporal order, based on the feature quantities for the respective first and second image samples.

21. The action recognition apparatus according to claim 19, wherein the similarity calculation unit calculates the similarity based on a ratio between probability densities of the first and second connected feature quantities.

22. A method for recognizing an action of an object in a plurality of time-sequential images, the method comprising:

extracting a first image sample from a plurality of time-sequential images belonging to a known action category;

extracting a second image sample from a plurality of time-sequential images that is a recognition target;

calculating feature quantities for the first image sample and feature quantities for the second image sample;

generating a first connected feature quantity and a second connected feature quantity by temporally connecting the feature quantities for the first image sample and the feature quantities for the second image sample, respectively;

calculating a similarity between the first connected feature quantity and the second connected feature quantity; and

recognizing the action of the object based on the similarity.

23. A storage medium storing a program that causes a computer to execute the method according to claim 22.