A pyroelectric infrared sensor equiped with a multiple reflection optical device was used to receive the infrared radiation emitted from a human body. By collecting and analyzing the signatures of the analog signals outputted from the pyroelectric infrared sensor, the location information of the human body could be extracted. When a person was passing through the area to be observed at a different distance from the infrared sensing node, the maximum peak of the infrared signal would be decreased with the increasing of the distance from the person to the infrared sensing node. A novel method for target localization based on the maximum peak of infrared signals outputted from the pyroelectric infrared sensor was designed. Compared with the centroid localization algorithm, the maximum localization error of this method was reduced by 87.89 %.