Single-Pixel Imaging at High Pixel-Resolutions with Nonuniform Sampling Patterns

Indirect image measurement techniques called single-pixel imaging (SPI), since their introduction over a decade ago have led to a considerable amount of novel ideas about image measurement at various wavelength ranges, spectral imaging, imaging through scattering media, 3D imaging etc. Unfortunately due to the trade-offs imposed by the modulation bandwidth limitations and computational cost of image reconstructions, the usually reported pixel resolution of SPI varies between $32 \times 32$ and $256 \times 256$ pixels falling far below imaging standards with classical methods. We have recently proposed a SPI framework for imaging at the native resolution of digital micro-mirror devices, which involves a highly compressive measurement, and gives a reconstruction result with a quality that depends on image sparsity [Opt. Express 30, 22730, 2022]. In this approach, the sampling patterns used to probe the high-resolution image have a double role. On the one hand, they are necessary to examine information about the spatial composition of the image. Secondly, they allow us to determine empty areas in the image effectively. Subgroups of patterns partition the entire image in complete and redundant ways. We have previously proposed a single partitioning example. In the present paper, we introduce another partitioning that focuses on specific areas of the image. The reconstruction quality of these regions of interest is enhanced.