An image processing device includes an input unit which inputs ordinary frames in a state in which an object is irradiated with ordinary light, and a special frame in a state in which the object is irradiated with special light, which are imaged consecutively at a predetermined ratio according to a predetermined frame period; a detection unit which detects motion vectors of the object from a plurality of the ordinary frames with different imaging timing; a motion correction unit which subjects the special frame to motion correction corresponding to the imaging timing of the ordinary frames based on the detected motion vectors; and a compositing unit which subjects the ordinary frames to an image compositing process based on the special frame.