|M.Sc Student||Vadim Holodovsky|
|Subject||In-Situ Multi-View Multi-Scattering Stochastic Tomography|
|Department||Department of Electrical Engineering||Supervisor||Full Professor Schechner Yoav|
|Full Thesis text|
To recover the three dimensional (3D) volumetric matter distribution in an object, the object is imaged from multiple directions and locations. Using these images, tomographic computations seek the distribution. When scattering is significant and under constrained irradiance, tomography must explicitly account for off-axis scattering. Furthermore, tomographic recovery must function when imaging is done in-situ, as occurs in medical imaging and ground-based atmospheric sensing. When imaging is done on large scale, as in ground-based atmospheric sensing, tomographic recovery needs to rely on uncontrolled solar radiation. This differs from medical computational tomography (CT) where active illumination is used to probe the medium. We cannot rely only on direct-transmission.
We formulate an image formation model based on 3D radiative transfer, using a Monte-Carlo numerical method. With this model we formulate tomography as the solution to an optimization problem, trying to fit a model to the measurements. In our formulation the image formation model is highly parallelizable, which can enable large scale rendering and recovery of volumetric scenes with a large number of unknown variables. Our tomography handles arbitrary orders of scattering, and so does the image formation model. The complexity of light propagation in scattering media makes it difficult to invert the model and solve the proposed optimization problem. We perform the optimization using an efficient surrogate function approach. The in-situ setup causes instabilities which we tackle.