3D image reconstruction from sparse measurement of wideband millimeter wave SAR experiments

Nonuniform random sampling is applied to a wideband 3D synthetic aperture radar (SAR) imaging system to reduce the number of measurement points in both frequency and space domains. Experimental tests were performed on a construction foam specimen using uniformly sampled Q-band frequencies in 35.04 GHz to 44.96 GHz and at a grid of two millimeter step sizes. Using discrete Fourier transform (DFT) or discrete cosine transform (DCT) sparse representations, the 3D images can be reconstructed from 7% random samples of the experimental data achieving comparable quality as the one from original full-set data. This can translate to significant time reduction of measurement from more than one hour to less than 20 minutes.