To specify a subject position in an image in various scenes.

A subject specification apparatus includes luminance image/color difference image generation means which generates a luminance image and a color difference image on the basis of a target image, subject region extraction means which extracts a subject region including a subject from the generated luminance image and color difference image, representative value calculation means which calculates a representative value on the basis of pixel values of pixels within the subject regions extracted from the images, differential image generation means which subtracts the calculated representative value from the pixel value of each of the pixels in the subject region to create a differential image corresponding to each of the pixels, binarization means which binarizes the generated differential image, combination means which combines the binarized images, mask extraction means which divides the combination image into a plurality of images and extracting a group of white pixels from the divided images as a mask, evaluation value calculation means which calculates an evaluation value indicating that the possibility for the extracted mask to be a mask indicating the subject, and subject specification means which specifies the subject in the target image on the basis of the calculated evaluation value.