Method, medium, and system generating video abstract information

1. A method of generating a video abstract, comprising:

detecting, by one or more processors, a first section of video data based on audio information of the video data;

detecting a second section of the video data based on the first section, the second section being chronologically successive and including the first section;

detecting third sections, from the detected second section, based on visual information of the video data; and

generating video abstract information of the video data by using the detected third sections,

wherein the second section includes a time-wise forward section before the first section and a time-wise back section after the first section, the time-wise forward section has a predetermined time length and the time-wise back section has the predetermined time length.

2. The method of claim 1 , wherein, when the video data is associated with a sport game, the first section is determined based on one of cheers, a frequency of reaching a crescendo of an announcer, and a sound of a handclap.

3. The method of claim 1 , wherein, when the video data is associated with news, the first section is determined based on a silent section.

4. The method of claim 1 , wherein, when the video data is associated with one of a drama and a movie, the first section is determined based on one of a music section and an explosive sound.

5. The method of claim 1 , wherein the detecting of the third sections comprises:

detecting an event reference shot from the second section;

detecting an event construction shot time-wise adjacent to the detected event reference shot; and

determining a third event section based on the detected event construction shot.

6. The method of claim 5 , wherein, when the video data is associated with a soccer game, the event reference shot is a penalty area shot.

7. The method of claim 6 , wherein the detecting of the event reference shot comprises:

dividing a frame into N×N blocks and determining a first critical value with respect to a detected pixel value for each block;

allocating 255 to a pixel value for each respective block when the respective detected pixel value is greater than the first critical value;

allocating 0 to a pixel value for each respective block when the respective detected pixel value not greater than the first critical value:

substituting a position of a pixel whose allocated value is 255 through a predetermined equation and computing a position having a largest number of the pixels;

determining whether the computed position is corresponding to an incline and a distance of a penalty line when the largest number of the pixels is greater than a second critical value; and

detecting the event reference shot to be the penalty area shot when the computed position is corresponding to the incline and the distance of the penalty line.

8. The method of claim 6 , wherein the detecting of the event construction shot time-wise adjacent to the detected event reference shot comprises:

determining whether, in a following shot after the penalty area shot, a number of close-up shots is greater than a predetermined number; and

determining the close-up shots to be event construction shots when the number of close-up shots is greater than the predetermined number.

9. The method of claim 8 , wherein, when the video data is associated with the soccer game, in the determining of the third section, the penalty area shot and the close-up shots are determined to be part of the third section.

10. The method of claim 5 , wherein, when the video data is associated with news, the event reference shot is an anchor shot.

11. The method of claim 10 , wherein the detecting of the event reference shot comprises:

extracting an anchor model candidate shot;

generating an anchor model from the extracted anchor model candidate shot; and

determining the event reference shot to be the anchor shot by comparing the generated anchor model with a key frame.

12. The method of claim 5 , wherein, when the video data is associated with a movie, the event reference shot is a front close-up shot, and the event construction shot is an action shot.

13. The method of claim 12 , wherein the detecting of the event reference shot comprises:

detecting a face from a shot of the video data;

measuring a distance between eyes of the detected face; and

determining the event reference shot to be the front close-up shot when the measured distance between the eyes is greater than a critical value.

14. The method of claim 1 , wherein the detecting of the second section in the video data is based only on the corresponding audio information.

15. The method of claim 1 , wherein the generating of the video abstract information is based only on third sections that are derived from the detected second sections based on the audio information.

16. A non-transitory medium comprising computer readable code to implement the method of claim 1 .

17. An apparatus, comprising:

a section detection unit to detect a first section of video data based on audio information of the video data, to detect a second section of the video data based on the first section, the second section being chronologically successive and including the first section, and to detect third sections, from the detected second section, based on visual information of the video data; and

a video abstract information generation unit to generate video abstract information of the video data by using the detected third sections,

wherein the second section includes a time-wise forward section before the first section and a time-wise back section after the first segment, the time-wise forward section has a predetermined time length and the time-wise back section has the predetermined time length.

18. The method of claim 1 , wherein the second section includes at least one of a time-wise forward section immediately before the first section and a time-wise back section immediately after the first segment.

19. The method of claim 1 , a start time of the second section is earlier than a start time of the first section, and a finish time of the second section is later than a finish time of the first section.