Image identification device and program

[Problem] To highly precisely identify the category to which an input image belongs even when extraction of the subject of identification is difficult or when the shape, size, orientation, position, and the like of the subject of identification are diverse/irregular. [Solution] The present invention has: a first feature extraction unit that divides an input image into a plurality of block images of a predetermined size, and determines a feature quantity of each of the plurality of block images; a region splitting unit that, on the basis of the feature quantities, allocates each of the plurality of block images to one of K (K being an integer that is at least 2) regions, and splits the input image into the K regions; a second feature extraction unit that, on the basis of the region splitting results of the region splitting unit, determines the feature vector containing as an element the proportion of block images allocated to each of the K regions among the plurality of block images or the number of the plurality of block images allocated to each region; and an identification unit that, on the basis of the feature vector, identifies which of a plurality of categories the input image belongs to.