We present a novel and fast metod for utilizing wavefront information in closed-loop phase-diverse image data. We form a 2-dimensional object-independent error function using the images at different focus positions together with OTFs of the diffraction limited system. Each coefficient in an expansion of the wavefront is estimated quickly and independently by calculating the inner product of a corresponding predictor function and the error function. This operation is easy to parallelize. The main computational burden si in pre-processing, when the predictors are formed. This makes this method fast and therefore attractive for closed loop operation. Calculating the predictors involves error function derivatives with respect tot he wavefront parameters, statistics of the parameters, noise levels and other known characteristics of the optical system. The predictors are optiized so that the RMS error in the wavefront parameters is minimized rather than consistency between estimated quantities with image data. We present simulation results that are relevant to the phasing of segmented mirrors in a space telescope, such as the NGST.
展开▼
