Time-coded aperture for x-ray imaging

We describe a computational imaging system for x-ray tomography, where image capture and post-processing are co-designed to improve final image quality when relative motion of an experiment's components during a single exposure causes motion blur. The idea is based on temporally encoding the motion during each exposure by fluttering the detector shutter open and closed with a known sequence for guaranteeing an invertible motion blur kernel. While generally applicable, we demonstrate our approach by simulating blurry data acquisition for transmission x-ray tomography and deblurring the reconstructed images. The results suggest that optimized pseudo-random binary time-coded apertures can yield successful reconstructions independent of the size of the blur kernel. This Letter is especially relevant to high-speed x-ray tomography applications where time-resolution is limited by the detector or available photon flux. (c) 2019 Optical Society of America