Superresolution of Passive Millimeter-Wave Imaging

摘要

This project develops a new method for image acquisition to improve resolution. Ordinarily, a focal plane sensor array is arranged in a rectangular grid at sub-Nyquist spacing, and the array must be dithered to sample the image plane at the Nyquist rate in each dimension. However, the Nyquist rate oversamples the image due to the usually circular support of the diffraction- limited image spectrum. We develop efficient algorithms for optimizing the dithering pattern so that the image can be reconstructed as reliably as possible from a periodic nonuniform set of samples obtained from a dithered rectangular- grid array. Furthermore, we develop efficient reconstruction algorithms that can quickly reconstruct the uniformly sampled image from its nonuniform samples. To support this framework, we derive a reconstruction method that accommodates the shift-variant effects of boundary conditions as well as the effect arising from the need to smooth the image less near edges to reconstruct sharp edges from low-resolution data. The method decomposes the image into two parts, one of which can be computed with FFT's and the other requiring a small matrix inversion. The resulting methods allow faster image acquisition as well as faster reconstructions that contain less noise and fewer artifacts when restoring and superresolving the acquired images.