Direct millimeter-wave (MMW) holographic imaging, which provides both the amplitude and phase information by using the heterodyne mixing technique, is considered a powerful tool for personnel security surveillance. However, MWW imaging systems usually suffer from the problem of high cost or relatively long data acquisition periods for array or single-pixel systems. In this paper, compressive sensing (CS), which aims at sparse sampling, is extended to direct MMW holographic imaging for reducing the number of antenna units or the data acquisition time. First, following the scalar diffraction theory, an exact derivation of the direct MMW holographic reconstruction is presented. Then, CS reconstruction strategies for complex-valued MMW images are introduced based on the derived reconstruction formula. To pursue the applicability for near-field MMW imaging and more complicated imaging targets, three sparsity bases, including total variance, wavelet, and curvelet, are evaluated for the CS reconstruction of MMW images. We also discuss different sampling patterns for single-pixel, linear array and two-dimensional array MMW imaging systems. Both simulations and experiments demonstrate the feasibility of recovering MMW images from measurements at 1/2 or even 1/4 of the Nyquist rate.

No abstract

Millimeter-wave holographic imaging system can acquire 3D images and is nonionzing, which especially fits human imaging. Widely utilized linear antenna arrays provide high scanning speed as well as non-uniformity artifacts. In this paper, we developed a set of calibration methods and image denoising algorithms to eliminate the influences of system errors including response non-uniformity, transmission delay and background scattering. The experimental results prove that the calibration methods contribute to non-uniformity reduction. The denoising algorithm is efficient in eliminating the disturbances of artifacts related to sub-sampling.