An image processing device includes: an extractor configured to extract a region of interest which includes a point of interest and satisfies a specified condition in a first image frame; a divider configured to divide the region of interest into a first subregion including the point of interest and a second subregion not including the point of interest at a narrow portion of the region of interest; and a specifying unit configured to specify a specified pixel in the first subregion as a point of interest of a second image frame.
