An image obtained by imaging a subject is displayed and a tracking frame is displayed at the central portion of a display screen. A target area is set surrounding the tracking frame and a high-frequency-component image is generated. A distance image indicating the distance to the subject image within an imaging zone is generated. An area, which represents a subject at a distance identical with that of the subject portion specified by the tracking frame displayed at the reference position, is decided upon as a search area. While a moving frame is moved within the search area of the high-frequency-component image, amounts of high-frequency component are calculated. The position of the moving frame at which the calculated amount of high-frequency component is maximized is adopted as the initial position of the tracking frame.