Automatic extraction of secondary video streams

1. A method of automatically generating at least one secondary video stream based on a primary video stream, performed by a computer responsive to a user interface, the method comprising:

performing video analysis on the primary video stream to generate one or more analysis results, the one or more analysis results including information corresponding to at least one of a target of interest, an event of interest, and an area of interest;

detecting, by the computer:

foreground objects in the primary video stream, and

at least one of the target of interest, the event of interest, and the area of interest using the one or more analysis results and at least one of an automatically detected criterion and a user specified criterion received by the computer via the user interface;

classifying, using a neural network, at least one of the detected target of interest and the detected event of interest; and

in response to the classifying, automatically extracting a first secondary video stream having a virtual field of view defined to be smaller than a field of view of the primary video stream,

wherein the first secondary video stream comprises at least a portion of at least one of the detected target of interest and the detected event of interest.

2. The method of claim 1 ,

wherein the primary video stream is a higher resolution primary video stream, and

wherein the performing the video analysis comprises:

down-sampling the higher resolution primary video stream to obtain a lower resolution primary video stream; and

performing the video analysis on the lower resolution primary video stream.

3. The method of claim 2 , further comprising:

extracting the first secondary video stream from the higher resolution primary video stream.

4. The method of claim 1 , wherein the detecting the foreground objects in the primary video stream comprises:

detecting the foreground objects by determining regions in the primary video stream containing newly-modified objects.

5. The method of claim 1 , further comprising:

automatically extracting, as the first secondary video stream, a best-shot of the detected target of interest in the primary video stream.

6. The method of claim 1 , further comprising:

automatically extracting, as the first secondary video stream, a face of a human target detected in the primary video stream.

7. The method of claim 1 , further comprising at least one of:

classifying the detected target of interest as a human target, and extracting the first secondary video stream comprising a face of the human target; and

classifying the detected target of interest as the human target, and extracting the first secondary video stream comprising an upper portion of the human target.

8. The method of claim 1 , further comprising:

classifying the detected target of interest as a vehicle target, and extracting the first secondary video stream comprising a license plate of the vehicle target.

9. The method of claim 1 , wherein the virtual field of view of the first secondary video stream is a portion of the field of view of the primary video stream.

10. The method of claim 1 , wherein the one or more analysis results comprise information including one or more of data describing a location of the detected target of interest in relation to the primary video stream, data describing a location of the detected target of interest, data describing a location of a base of a detected object, and data describing at least one of the detected target of interest and the detected event of interest.

11. The method of claim 1 , further comprising:

determining an orientation of the detected target of interest in the primary video stream; and

automatically extracting the first secondary video stream in response to the determined orientation of the detected target of interest.

12. The method of claim 1 , further comprising:

determining a direction of motion of the detected target of interest in the primary video stream; and

automatically extracting the first secondary video stream in response to the determined direction of motion of the detected target of interest.

13. The method of claim 1 , further comprising:

automatically extracting a second secondary video stream,

wherein the first secondary video stream is extracted in response to detecting a first target and the second secondary video stream is extracted in response to detecting a second target.

14. The method of claim 1 ,

wherein the performing the video analysis comprises analyzing the primary video stream to detect an object that has been left behind, and

wherein the automatically extracting the first secondary video stream comprises automatically extracting the first secondary video stream including the object that has been left behind.

15. The method of claim 1 , further comprising:

calculating a three-dimensional (3D) position for the detected target of interest.

16. The method of claim 1 , further comprising:

tracking the detected target of interest using the first secondary video stream.

17. The method of claim 1 , wherein the automatically extracting the first secondary video stream comprises at least one of:

automatically extracting the first secondary video stream from a predefined location in the primary video stream; and

automatically extracting the first secondary video stream based upon a predefined area in the primary video stream.

18. The method of claim 1 , wherein the automatically extracting the first secondary video stream comprises:

automatically extracting the first secondary video stream with a predefined size.

19. The method of claim 1 , further comprising:

analyzing the first secondary video stream to detect a first target of interest; and automatically extracting a second secondary video stream in response to detecting the first target of interest.

20. The method of claim 1 , further comprising:

identifying one or more foreground portions in the primary video stream;

grouping the one or more foreground portions into one or more targets of interest; and

generating at least one video primitive describing at least one characteristic of at least one of said one or more targets of interest.