A feature amount acquisition unit divides input image data into blocks each including a plurality of pixels and acquires the feature amounts of the blocks. A first detection unit detects a pair of two blocks whose feature amounts are approximate to each other and extract a first aggregate area in which one block of each of the pairs gathers and a second aggregate area in which the other block of each of the pairs gathers from the input image data. A second detection unit performs additional detection of adding a block adjacent to the first aggregate area to the first aggregate area, and adding a block adjacent to the second aggregate area to the second aggregate area.