An image processing device in which a high-resolution viewpoint image and a disparity map is obtained with a simple configuration. A resolution of a viewpoint image imaged at a central viewpoint at first resolution is converted to a resolution of a viewpoint image imaged at a peripheral viewpoint at second resolution which is lower than the first resolution, and the viewpoint image at the central viewpoint and the viewpoint image at the peripheral viewpoint both having a same resolution are generated. Further, a disparity map based on a correlation value between the viewpoint image at the central viewpoint and the viewpoint image at the peripheral viewpoint both having the same resolution is generated.