To obtain an image having high level of satisfaction even when facial expressions of a plurality of image-captured people are different.

When an automatic imaging mode is set to "ON", a face detection unit detects four human faces from an image of image data from an A/D conversion unit, and supplies a facial expression evaluation unit with information such as positions and sizes of the detected faces. The facial expression evaluation unit calculates a facial expression evaluation value for each person as illustrated in C to G with dotted lines. A threshold value control unit changes standards by lowering by one step a threshold value of a crying face, which the smallest number of people have, on the basis of the facial expression evaluation values and the number of people with each facial expression in the image. When each of the facial expression evaluation values exceeds a corresponding threshold value, the image is extracted. This disclosure is applicable, e.g., to an imaging apparatus which captures an object.