A hairstyle simulation image is formed by combining a face image and a hair image such that a hair image of any hairstyle can be simply combined with the face image without impairing the naturalness of the fitted face and hair images, irrespective of whether or not a part of the face is covered with the hair. The hairstyle simulation image formed by combining the face and hair images is formed as follows. First, two eyes in the face image are detected, and a semi-oval shaped curve L0 is formed based on a facial width D0 and a line passing through the eyes (an eye line L1). The facial width D0 is obtained by multiplying the distance between the eyes by a predetermined multiplier, and the semi-oval shaped curve L0 serves as a temporary facial outline. By moving the semi-oval shaped curve L0 in the right and left directions on the face image, the left and right edges Pb and Pc of the facial outline of the face image are detected based on the degree of overlap between the semi-oval shaped curve L0 and an edge line La of the jaw portion of the face image. The left and right edges Pb and Pc of the outline of the hair image are also detected as in the above. The widths D1f and D1h between the left and right edges Pb and Pc of the facial outlines of the face and hair images, respectively, are adjusted to the same width, and then the face and hair images are combined with each other.