Detection of trapped survivors under collapsed buildings using impulse-radio ultra-wideband (IR-UWB) radar has become an important research topic. However, due to the jitter or drift of the radar, clutter reflected from walls or rubble is similar to the human respiratory response in some scenarios. Thus, it becomes difficult to determine the presence or absence of the survivor under ruins. In this paper, a novel dual-frequency IR-UWB radar with low center frequencies is developed and a method based on adaptive clutter cancellation is proposed to eliminate the respiration-like clutter. Experiments are carried out in both through-wall and simulated rubble cases, and the results shows that the proposed method outperforms the common method, which is based on linear-trend subtraction with single-frequency IR-UWB radar in removing the respiration-like clutter and improves the radar performance in life detection.