To generating a mosaic image by combining a plurality of material images, an original image is divided into tiles and a material image having a characteristic similar to that of an image in each tile is applied to the tile. If more than one version of image data having different resolutions is provided as the image data for the material image, a low-resolution version of the image data is used to calculate a characteristic quantity of the image. Thus, the amount of time required to generate the mosaic image can be reduced.