Assessing correction accuracy in image-based adaptive optics

Modal sensorless adaptive optics relies on the use of an image quality metric to estimate the amplitude of aberrations, and of a well-suited set of aberration modes to describe the aberration. This set is chosen so that aberration of one mode does not influence correction in another mode. In this paper, we show how these modes can be derived experimentally, and investigate the influence of imperfect crosstalk removal on the accuracy of correction. We show that the resulting error can be mitigated using appropriate algorithms that can incorporate knowledge of the influence of the modes on the metric and, if available, partial knowledge of the aberrations. Finally, we derive from these results the minimum time required for correction in various situations.