On a play field 20 set in real space, a target region 102 having the play field 20 as a bottom surface and having a predetermined height h is set virtually. A target region image that expresses a distance of the target region 102 from an imaging surface as a pixel value on an image plane is generated, and compared with a depth image corresponding to an actual captured image 100; thus, only images of blocks 3c and 3b inside the target region 102 are extracted as detection and tracking targets.