Deep iterative reconstruction for phase retrieval

Isil, Cagatay
Öktem, Sevinç Figen
Koc, Aykut
The classical phase retrieval problem is the recovery of a constrained image from the magnitude of its Fourier transform. Although there are several well-known phase retrieval algorithms, including the hybrid input-output (HIO) method, the reconstruction performance is generally sensitive to initialization and measurement noise. Recently, deep neural networks (DNNs) have been shown to provide state-of-the-art performance in solving several inverse problems such as denoising, deconvolution, and superresolution. In this work, we develop a phase retrieval algorithm that utilizes two DNNs together with the model-based 1-110 method. First, a DNN is trained to remove the HIO artifacts, and is used iteratively with the HIO method to improve the reconstructions. After this iterative phase, a second DNN is trained to remove the remaining artifacts. Numerical results demonstrate the effectiveness of our approach, which has little additional computational cost compared to the HIO method. Our approach not only achieves state-of-the-art reconstruction performance but also is more robust to different initialization and noise levels. (C) 2019 Optical Society of America


Geometric measures of entanglement
UYANIK, KIVANÇ; Turgut, Sadi (American Physical Society (APS), 2010-03-01)
The geometric measure of entanglement, which expresses the minimum distance to product states, has been generalized to distances to sets that remain invariant under the stochastic reducibility relation. For each such set, an associated entanglement monotone can be defined. The explicit analytical forms of these measures are obtained for bipartite entangled states. Moreover, the three-qubit case is discussed and it is argued that the distance to the W states is a new monotone.
Exact Pseudospin Symmetric Solution of the Dirac Equation for Pseudoharmonic Potential in the Presence of Tensor Potential
AYDOĞDU, OKTAY; Sever, Ramazan (Springer Science and Business Media LLC, 2010-04-01)
Under the pseudospin symmetry, we obtain exact solution of the Dirac equation for the pseudoharmonic potential in the presence of the tensor potential with arbitrary spin-orbit coupling quantum number kappa. The energy eigenvalue equation of the Dirac particles is found and the corresponding radial wave functions are presented in terms of confluent hypergeometric functions. We investigate the tensor potential dependence of the energy of the each state in the pseudospin doublet. It is shown that degeneracy b...
IKHDAIR, SM; Sever, Ramazan (Springer Science and Business Media LLC, 1993-09-01)
The energy eigenvalues of bound states of an electron in the general exponential cosine screened Coulomb potential are obtained using the shifted 1/N expansion method. The energies for the states from 1s to 8k are calculated from six to eight significant figures. The energy eigenvalues for the 1s, 2s - 2p, 3s - 3d, and 4s - 4f states are also presented as a function of the screening parameter lambda. Results are compared with the ones obtained by other workers. The agreement reduces roughly for large lambda...
Strong oscillations in the nondipole corrections to the photoelectron angular distributions from C-60
Toffolı, Danıele; Decleva, Piero (American Physical Society (APS), 2010-06-25)
Nondipolar corrections to the photoelectron angular distributions from C-60 have been calculated for the highest occupied molecular orbital ( HOMO), HOMO-1, and HOMO-2 photoemission bands. The computational method employed takes advantage of a parallel algorithm that uses a multicentric expansion of bound- and scattering-wave functions and a density-functional theory one-particle Hamiltonian. First-order nondipolar asymmetry parameters have been calculated from thresholds of up to 160 eV of photon energy. S...
One analytic form for four branches of the ABCD matrix
Baskal, S.; Kim, Y. S. (Informa UK Limited, 2010-01-01)
It is not always possible to diagonalize the optical ABCD matrix, but it can be brought into one of the four Wigner matrices by a similarity transformation. It is shown that the four Wigner matrices can be combined into one matrix with four branches. This result is illustrated in terms of optical activities, laser cavities, and multilayer optics.
Citation Formats
C. Isil, S. F. Öktem, and A. Koc, “Deep iterative reconstruction for phase retrieval,” APPLIED OPTICS, pp. 5422–5431, 2019, Accessed: 00, 2020. [Online]. Available: