Three-dimensional image reconstruction. I. Determination of pattern orientation

The problem of determining the Euler angles of a randomly oriented three-dimensional (3D) object from its 2D Fraunhofer diffraction patterns is discussed. This problem arises in the reconstruction of a positive semidefinite 3D object using oversampling techniques. In such a problem, the data consist of a measured set of magnitudes from 2D tomographic images of the object at several unknown orientations. After the orientation angles are determined, the object itself can then be reconstructed by a variety of methods using oversampling, the magnitude data from the 2D images, physical constraints on the image, and then iteration to determine the phases.