In this paper a novel change detection technique is introduced and used to process modeling and measurement data of a person walking inside a building. The proposed change detection technique is based on non-coherent change detection. The MT signature, back-wall shadow, front-wall response, and sidelobe-artifact suppression generated by the proposed algorithm are analyzed and compared with those generated by coherent change detection. It will be shown that the proposed algorithm greatly attenuates imaging artifacts while preserving the MT signature.