Digital Holograms Obtained Using Advanced Evolution Strategies

Abstract

By changing the amplitude and phase of optical beams digital holography allows for the generation of desired target images at the focus of a lens. It is an important research challenge to determine the digital hologram (especially the phase function) required to generate a specific target image in a fast and precise manner, especially in dynamically changing environments. In this study, the digital hologram required to obtain a certain target image was calculated with two different algorithms and another hybrid approach combining both algorithms, and the obtained results were compared. Gerchberg Saxton (GS) and CMA-ES (Covariance Matrix Adaptation Evolution Strategy) algorithms were used separately and in a hybrid manner. GS algorithm is a wellestablished inverse Fourier Transformation based algorithm for digital hologram calculation. Our study shows the suitability of the CMA-ES algorithm for this problem. Although the CMA-ES algorithm requires a high number of iterations and results in a slower implementation, it provides a more general approach than the GS algorithm and it allows for the use of different fitness functions that are not necessarily based on Fourier Transformation. In future studies, it is planned to parallelize the execution of the CMA-ES algorithm using GPUs and use advanced versions of the CMA-ES algorithm to improve the performance in higher dimensional problems.