A Fast Back-Projection Approach to Diffraction Tomography for Near-Field Microwave Imaging

A novel fast back-projection (FBP) approach is presented as a more robust alternative to conventional diffraction tomography (DT) for near-field three-dimensional (3-D) microwave imaging. Like DT, the FBP algorithm utilizes the spectral Green's function and stationary phase solution in the forward model, enabling the implementation of the fast Fourier transform to achieve real-time image generation with the same efficiency as DT. However, if the Green's function or the antenna gain pattern has any nulls within the imaging volume, the DT can have singularities that give rise to unbalanced images. The FBP not only naturally avoids the singularity problem and obtains balanced images through the point spread function, but is also much less sensitive to noise due to the matched-filter property of back-projection. 3-D simulations and measurement of a monostatic near-field free-space imaging scenario are presented to demonstrate the efficiency of the proposed FBP method in comparison with DT, and the improved uniformity and noise resistance of the image.