An image processing device includes a reference camera determiner that determines one among the multiple cameras, as a reference camera; an imaging condition acquisition unit that acquires imaging conditions relating to exposure and white balance which are set in the reference camera; an imaging condition setter that sends a command to set the imaging conditions relating to the exposure and the white balance, based on the imaging conditions in the reference camera, to cameras other than the reference camera; and an image composition unit that performs image composition processing on the multiple captured images that are acquired from the multiple cameras and that outputs the panoramic composite image. Thus, a boundary of an area of each composed image in the panoramic composite image that results from the composition is suppressed from being unnaturally conspicuous, and the captured image that is a reference is suitably displayed.