Optimal Zernike Parameterization in Global Phase Retrieval of Large Reflector Antennas

Phase retrieval is a widely-used technique to reconstruct the surface deformations of large reflector antennas. To cope with the local stagnation of the classical phase retrieval algorithm, we previously proposed a hybrid global algorithm named CMAES-HIO, in which the Zernike polynomials are first used to parameterize the aperture initial phase. This paper investigates the effect of the Zernike order on the algorithm performance. By introducing a hybrid evaluation parameter combining algorithm accuracy and time consumption, the effect of Zernike order on algorithm performance is investigated for the cases of gravitational and simulated random continuous deformations. The results show that, for the cases studied, which are representative of a wide class of deformations, the Zernike order in the range of 5 or 6 is optimum for the comprehensive performance of algorithm with respect to time and accuracy.