The person extraction unit 10 analyzes the moving image data 100 to be analyzed for each frame, and detects a person(s) from each frame. Since a unit that detects a person(s) from an image can be implemented according to the related art, explanation thereof will not be repeated herein. The person extraction unit 10 may analyze all frames as data to be processed and perform person detection, or may intermittently analyze frames as data to be processed (for example, every M frames, provided M is an integer of 1 or more).

The person extraction unit 10 assigns a detection ID to each person detected from each frame. In a case where the same person is detected from a plurality of frames, a detection ID is provided for each detection from each frame.

Then, as shown in FIG. 5, the person extraction unit 10 associates predetermined information with the detection ID. First, the person extraction unit 10 extracts an outer appearance feature (for example, a face feature) from the image of each person in each frame, and associates the extracted feature value with the detection ID. In addition, the person extraction unit 10 associates a frame ID, which indicates the frame the person was detected in, with the detection ID. The person extraction unit 10 may further generate a file of the face image of each person in each frame, and associate the file with the detection ID. The file of a face image is generated by cutting out a portion (a face portion of each person) from each frame.