Complexity-Guided Phase Retrieval Algorithm

The popular iterative phase retrieval algorithms like Gerchberg-Saxton (GS), Fienup HIO and their variants suffer from twin-image stagnation problem and take a very large number of iterations for their convergence. A novel approach ‘complexity-guided phase retrieval (CGPR)’ for the complex-object recovery is presented in this paper. We show that a new metric called ‘complexity parameter’ which provides information about the object can be directly calculated from the Fourier intensity data. The CGPR algorithm shows high-quality artifact-free phase reconstruction in a significantly less number of iterations.