An ocular fundus image is captured by a fundus camera (101). Blood vessel regions (12) are erased from the ocular fundus image, and a luminance region greater than a prescribed luminance threshold value (304) and having an area equivalent to a predetermined standard disc area is extracted as a region (11) having a standard disc area. The region (11) having a standard disc area is divided into a plurality of regions (S1 through S8) by a plurality of dividing lines (14a through 14h) extending radially from a center of gravity (13) thereof, and divided regions (S1 through S8) are scanned radially from the center of gravity while angles are shifted to detect in each individual region a point (15a) at which luminance variation reaches a maximum. The contour line (501) of the disc region (500) is derived from these points (15a through 15h).