IST Lunch Bunch

Abstract

Three-dimensional (3D) volumetric imaging is traditionally performed in a highly controlled environment, e.g. under a static microscope or in an x-ray gantry. We derive generalized approaches for 3D volumetric recovery in uncontrolled, random settings, suitable to natural objects. First, we leverage scattering effects, to achieve passive tomography of large objects (clouds, atmosphere), where the radiation source (Sun) cannot be controlled. Furthermore, multi-view image data leads to the retrieval of the microphysical properties of scatterers within the medium. Then, we look at statistical tomography: volumetric 3D recovery based on a random population of specimens, where each specimen in the population is viewed only once, at random unknown scale, orientation and location. This generalization, drawing from Cryo-EM single particle reconstruction, is useful for the study of microscopic and mesoscopic organisms, e.g. plankton.