To efficiently extract a best shot image while recording only the image (the best shot image), in which a face that a user desires to record is photographed, from among the continuously photographed images.

Continuously photographed frame images are captured, and temporarily stored (step S12). Feature points of a face are extracted from the captured frame images (step S14). A sum (an expression change amount) of distances between the feature points of the face (face parts) in preceding/following frame images is calculated (step S16). A number of m preceding/following frame images including the target frame image in which the expression change amount is the largest are extracted as best image candidates (step S20). A best shot image is selected from among the best image candidates so as to record the best shot image in a recording medium (step S22).