Generalized Phase Diversity Wavefront Sensing Based on Stochastic Parallel Optimization Algorithm

Main methods on phase retrieval of GPD (Generalized Phase Diversity) wavefront sensor are based on GS (Gerchberg-Saxton) iteration or its derivations, which do not make use of peculiarities of the sensor adequately and have heavy computation. We first decompose the output signal of sensor by Zernike modes and then use SPGD (Stochastic Parallel Gradient Descent) to search the optimum solution. Two different phase screens are used to investigate the phase retrieval method. Simulation results show the method we put forward can correctly restore the phase from the output signal of the GPD wavefront sensor for randomly generated distorted wavefronts and is easy to implement because Zernike modes can be obtained or calculated in advance, which offers the theory basis for further research on GPD wavefront sensor and its real applications.