Disclosed is a technique that eliminates problems that result when a face image fails to be detected in a case where the image of a subject obtained continuously is subjected to face-image detection processing. A face-image portion is detected in the image of a subject. If an evaluation value for evaluating the degree of face likeliness of the face-image portion is equal to or greater than a threshold value, the result of face detection is updated. A timer is set. If the timer has not timed out in a case where the evaluation value of a face image in the next frame of the image of the subject is less than the threshold value, the face-image portion of the preceding frame is regarded as the face-image portion of the next frame and processing regarding this face-image portion is executed. Thus, even if a face-image portion is no longer detected, processing regarding a face-image portion can be executed using the face-image portion of the preceding frame.