This paper proposes an approach to extract focused objects (i.e., attention objects) from low depth-of-field images. To recognize the focused object, we decompose the image into multiple regions, which are described by using three types of visual des