Background Removal for the Processing of Scans Acquired with the, “UGO-1st”, Landmine Detection Platform

Field tests of the robotic platform “UGO-1st” have revealed interference sources that complicate the automatic detection of subsurface objects by UWBGPR. Although robust target discrimination criteria make it possible to reduce the probability of false alarms, identification of these sources of interference improves the detection algorithm. The focus of this work is on the elimination of interference sources, which include: direct coupling between radar antennas; reflections from components of the robotic platform; reflections from the soil surface; instability of the GPR signal due to temperature drift during a scan. To reduce these effects we performed measurements with the robotic system suspended in air high above the ground so that signals from any subsurface reflection would arrive long after the signals from the robot structure that are to be identified and removed. These signals are also used to compensate for temperature drift, as well as to remove signals from direct coupling between antennas. The procedure for the background removal includes: selection of a time window before the reflection from surface; an optimization procedure for determining the time drift caused by temperature changes; alignment of the two signals; and subtraction of the background signal from the received one. This is performed for all four receiving channels separately. The need to remove reflections from the soil surface is due to the fact that these signals arrive at the receiver almost simultaneously with the reflections from shallow buried objects. This makes it difficult to determine the arrival time of the reflected signal. Reflections from the soil surface can also be eliminated in a manner similar to the one mentioned above but instead of measuring a background signal we took signals collected a few cm ahead of a known target. The algorithm was applied to real data and has proven to be effective.