Data Collection and Analysis for Personnel Detection at a Border Crossing

摘要

There is considerable interest in detecting people crossing the border with fewer false alarms and high confidence. This capability requires understanding the phenomenology of various sensor modalities and developing algorithms based on the phenomenology. In an effort to develop this capability, U.S. Army Research Laboratory scientists went to the southwest border to collect data using acoustic, seismic, passive infrared (IR), profiling, electric field, magnetic field, radar, sonar, visible, and IR imaging sensors. In this report, we discuss the data collection effort and resultant data, phenomenology of various sensor modalities, and robust detection algorithms. In the future, the acoustic sensor data will be processed to determine the characteristic features of human voice (formants, etc.), the seismic data will be processed using the ground transfer function to determine the cadence of the person walking (as opposed to an animal), and the radar and ultrasonic data will be processed to determine the Doppler frequency resulting from various limb movements.