Statistical Tomography

We achieve tomography of 3D volumetric natural objects, where each projected 2D image corresponds to a different specimen. Each specimen has unknown random 3D orientation, location, and scale. This imaging scenario is relevant to microscopic and mesoscopic organisms, aerosols and hydrosols viewed naturally in a microscope. In-class scale variation inhibits prior single-particle reconstruction methods. We thus generalize tomographic recovery to account for all degrees of freedom of a similarity transformation. This enables geometric self-calibration in the imaging of transparent objects. We make the computational load manageable and reach good quality reconstruction in a short time. This enables the extraction of statistics that are important for a scientific study of specimen populations, specifically size distribution parameters.


We apply the method to study plankton. Below are plankton images taken from Woods Hole Oceanographic Institution. The orientation and scale of similar plankton samples are recovered along with a statistical 3D density reconstruction. 

Very nice follow-up work by Roi Ronen et al. builds upon the problem formulation and framework we introduced and improves the algorithms and results. The improvement is achieved by introducing a statistical model for the noise sources: see the paper for details.


  1. Aviad Levis, Yoav Y. Schechner, Ronen Talmon, Statistical Tomography of Microscopic Life, Proc. IEEE CVPR, 2018, and Invited talk in CVPR Workshop on Automated Analysis of Marine Video for Environmental Monitoring (2018).