To obtain a shot image which can be recognized as a stereoscopic image without using a dedicated display device, and to have the obtained shot image recognized as a stereoscopic image.

A plurality of subject portions in a shot image are grouped by distance range in depth direction in a real space, and are divided into as many divided images as the number of groups, each consisting of only the subject portions belonging to the same group. The divided images are each enlarged so that each corresponding group will have such a positional relation that the subject portions of the divided image on the near side become larger than the subject portions of the divided image on the deep side, and layer structures reflecting the positional relations of each corresponding group are reserved and are combined by overwriting to generate a display image G, which is displayed. In the meantime, the subject portions in each layer except the lowermost layer are moved together in the crosswise direction in such a way that the subject portions on relatively upper layer side are moved a greater distance than are the subject portions on relatively lower layer side.