An adaptive multi-sensor visual attention model