Sparse sampling and enhanced axial resolution in millimeter-wave holographic imaging

This paper describes an active millimeter-wave (MMW) holographic imaging system used for the study of compressive measurement for concealed weapons detection. We record a digitized on-axis, Gabor hologram using a single pixel incoherent receiver that is translated at the detector plane to form an image composite. Capturing measurements in the MMW regime can be costly since receiver circuits are expensive and scanning systems can be plagued by their long data acquisition times. Thus, we leverage recent advances in compressive sensing with a traditional holographic method in order to estimate a 3D (x,y,z) object distribution from a 2D recorded image composite. To do this, we minimize a convex quadratic function using total variation (TV) regularization. Gabor holograms are recorded of semi-transparent objects, in the MMW, mimicking weapons and other objects. We present preliminary results of 3D reconstructions of objects at various depths estimated from a 2D recorded hologram. We compare backpropagation results with our decompressive inference algorithm. A possible application includes remote concealed weapons detection at security checkpoints.