A center of rotation O of a camera 10 is determined on an axis 45 passing through the position between pickup target objects A, B and the camera 10 in the direction from a lens 3 to a CCD 2. Before rotation (in a first state), an optical axis 30 of the camera 10 swings to the left with the point O as center and with an angle e with respect to the axis 45, whereas after rotation (in the second state), the optical axis 30 swings to the right with the point O as center and with angle e with respect to the axis 45. The pickup target objects A, B, C form images on a pickup face 2a, respectively. In order that the formed image b.sub.2, for example, of the image picked up after movement may coincide with the formed image b.sub.1 of the image before movement, the image after rotation is moved in the cross direction. Then, the image before rotation and the image obtained by moving the image after rotation in the cross direction are input to a display which allows a stereovision due to binocular parallax, which are seen as a stereoscopic image.