Here, an example of the processing flow in which the person extraction unit 10 generates the detected person information shown in FIG. 6 will be described. This example is merely an example, and the present invention is not limited to those described herein.

The person extraction unit 10 analyzes one of a plurality of frames to be processed as a first frame to be processed. Then, when persons are detected from the frame, the person extraction unit 10 assigns detection IDs to all the detected persons. Then, as shown in FIG. 5, a feature value, a frame ID, a face image file ID, and the like are associated with each detection ID.

The person extraction unit 10 assigns a person ID in association with each of all the detected IDs, and registers the person IDs in the detected person information as shown in FIG. 6.

Thereafter, the person extraction unit 10 analyzes the next frame as a frame to be processed. Then, in a case where persons are detected from the frame to be processed, the person extraction unit 10 assigns detection IDs to all the detected persons. Then, as shown in FIG. 5, a feature value, a frame ID, a face image file ID, and the like are associated with each detection ID.

Thereafter, the person extraction unit 10 determines whether or not the outer appearance feature value of each person detected in the frame to be processed is similar by the predetermined level or more to that of any of persons detected in previously processed frames.