Discrete reconstruction of real phase objects: a comparison with computer-simulated phase objects

Abstract

The Gerchberg-Saxton algorithm for iterative phase retrieval is applied in reconstructing an optical wave arriving from a coherently illuminated phase object. Computer simulations show that the algorithm converges to the correct solution when magnitude information for both image and Fourier domains is provided. Phase retrieval from measured intensities in an experimental optical system is compared with the simulations. Charge-coupled-device sensors, in which the output voltage is a nonlinear function of the light intensity, are used to perform measurements of the quasi magnitude information. Because of the accuracy of the measured data, phase retrieval for real simple objects is possible, and schlieren images can be generated on the computer screen.