In order to extract a plurality of items of image data corresponding to respective ones of a plurality of documents from scan data more accurately, an image processing apparatus is provided with a detection unit adapted to detect a plurality of image areas corresponding to the plurality of documents; a setting unit adapted to set a reference area size from sizes of the plurality of image areas; a comparison unit adapted to compare the sizes of the plurality of image areas and the reference area size; and a decision unit adapted to decide that an image area detected by the detection unit has not been detected correctly if result of the comparison by the comparison unit exceeds a preset threshold value.