xi
3.2 Compressed sensing imaging results, (a) Magnitude of image
reconstructed by inverse Fourier transform using the full data set
(4096 uniformly sampled measurements), and (d) its phase. Note the
phase distortion inherent in the THz beam in (d). Compressed
sensing reconstruction result using 500 measurements (12%) from the
full data set: (b) magnitude, and (e) phase. Compressed sensing with
phase correction improves image quality and eliminates phase
distortion (see (c) and (f)). All figures show a zoom-in view on a 40
× 40 grid centered on the object..................... 25
3.3 Comparison of quality of image reconstruction between CS with and
without phase correction. As the number of measurements (M) used
in CS increases, the mean-squared error (MSE) between the
magnitudes of the reconstructed image and the reference image (see
Fig 3.2(a)), normalized by the energy of the reference image,
decreases. CS with phase correction shows superior reconstruction
performance................................ 27
3.4 Image reconstruction results using (a) compressive phase retrieval
(CPR) with the full data set (4096 magnitude measurements), and
(b) compressed sensing phase retrieval (CSPR) with a subset of 1500
measurements from the data set used in (a)............... 28
4.1 The THz compressive imaging setup. An approximately collimated
beam from the THz transmitter illuminates an object mask and is
partially (~50%) transmitted through a random pattern of opaque
pixels. The random patterns, the focusing lens and the receiver are
placed in order to most efficiently focus the THz beam onto the
receiver antenna. One complete time-domain waveform is collected
for each random pattern.......................... 33