Abstract
We are developing a cart-based system for the detection of trace explosives by active infrared backscatter hyperspectral imaging. Infrared quantum cascade lasers (QCL) are rapidly wavelength tuned from 6-11 microns to interrogate a surface while a focal plane array collects backscattered images. The resulting hyperspectral image cube is processed to detect and identify traces of explosives. The algorithm, trained on both measured and synthetic reflectance spectra, utilizes a convolutional neural network. We present data collected from sample coupons fabricated on relevant substrates (glass, painted metal, plastics) with trace mass loading levels at standoff distances of up to a few meters.