The objective of this work is to supress out-of-focus signal in multi-depth lensless holographic reconstructions from a single hologram. An expectation-maximization based framework with sparsity regularization and multiplicative gradient descent optimization has been developed. The comparison with the standard method of compressive holography (using TwIST algorithm) showed a better convergence speed. The axial resolution demonstrated was comparable to the diffraction limit.
See results here.
You can start studying Compressive Holographic Tomography with this paper.