Improved terahertz imaging with a sparse synthetic aperture array

Sparse arrays are highly attractive for implementing two-dimensional arrays, but come at the cost of degraded image quality. We demonstrate significantly improved performance by exploiting the coherent ultrawideband nature of single-cycle THz pulses. We compute two weighting factors to each time-delayed signal before final summation to form the reconstructed image. The first factor employs cross-correlation analysis to measure the degree of walk-off between time-delayed signals of neighboring elements. The second factor measures the spatial coherence of the time-delayed delayed signals. Synthetic aperture imaging experiments are performed with a THz time-domain system employing a mechanically scanned single transceiver element. Cross-sectional imaging of wire targets is performed with a one-dimensional sparse array with an inter-element spacing of 1.36 mm (over four lambda at 1 THz). The proposed image reconstruction technique improves image contrast by 15 dB, which is impressive considering the relatively few elements in the array. En-face imaging of a razor blade is also demonstrated with a 56 x 56 element two-dimensional array, showing reduced image artifacts with adaptive reconstruction. These encouraging results suggest that the proposed image reconstruction technique can be highly beneficial to the development of large area two-dimensional THz arrays.