| 摘要 |
The apparatus comprises a feature quantity extraction part for extracting a feature quantity of a subject from video captured by plural cameras, an In/Out point extraction part for extracting In/Out points indicating points in which a subject appears and disappears in each video captured, an In/Out region formation part for forming In/Out regions based on the In/Out points extracted, a correlation value calculation part for calculating a correlation value by obtaining the total sum of similarities every feature quantity of the subject in each of the plural combinations of In/Out points included in the In/Out regions, a frequency histogram creation part for creating a frequency histogram based on the correlation value, and a link relation information generation part for extracting a peak of the frequency histogram and estimating the presence or absence of a link relation between the plural cameras and generating link relation information. |
