Standoff trace detection of explosives with infrared hyperspectral imagery

In this work we present a hyperspectral image sensor based on MIR-laser backscattering spectroscopy for contactless detection of explosive substance traces. The spectroscopy system comprises a tunable Quantum Cascade Laser (QCL) with a tuning range of 7.5 μm to 9.5 μm as an illumination source and a high performance MCT camera for collecting the backscattered light. The resulting measurement data forms a hyperspectral image, where each pixel vector contains the backscattering spectrum of a specific location in the scene. The hyperspectral image data is analyzed for traces of target substances using a state of the art target detection algorithm (the Adaptive Matched Subspace Detector) together with an appropriate background extraction method. The technique is eye-safe and allows imaging detection of a large variety of explosive substances including PETN, RDX, TNT and Ammonium Nitrate. For short stand-off detection distances (<3 m), residues of explosives at an amount of just a few 10 μg, i.e. traces corresponding to a single fingerprint, could be detected. For larger concentration of explosives, stand-off detection over distances of up to 20 m has already been demonstrated.