An optical tomography system measures the light scattered by an object as a function of spatial coordinates and as a function of the illumination angle. The measured signals are digitally processed to produce a 3D image of the object. In this paper we describe how we can learn the shape of an object by constructing a neural network that models the optical system and training the network to match the experimentally measured data. The variables of the trained network yield the image of the unknown object at the end of training phase.
[1] Ulugbek, Papadopoulos, Shoreh, Goy, Vonesh, Unser, Psaltis, “A Learning Approach to Optical Tomography” Optica, May 2015.
|