Extensions of Compressed Imaging: Flying Sensor, Coded Mask, and Fast Decoding

In this paper, we outline some recent advances and identify some open problems in compressed sensing (CS) for remote imaging and related imaging applications. We propose new approaches to compressed remote sensing and image reconstruction, which exploit redundant/overlapping measurements (RMs), multiplexing imaging (MI), and an effective iterative procedure for decoding in the recovery phase. The RM approach involves the concept of data fusion and applies a noiselet transform to a CS measurement matrix, followed by 2-D hexagonal jittered sampling and 1-D jitter-based circular sampling. The MI approach enlarges the field of regard of imaging and provides the compromise between the numbers of masks and detectors by taking advantage of the merits of previous single-pixel sequent imaging and multipixel parallel imaging. Finally, the new decoding method in the CS recovery stage combines Bregman-based nonlocal total variation with curvelet-active-set iteration.