|M.Sc Student||Zommer Shahaf|
|Subject||Simulated Annealing in Adaptive Optics for Imaging the|
|Department||Department of Physics||Supervisors||Professor Emeritus Stephen Lipson|
|Dr. Joan Adler|
Adaptive optics is a method designed to correct images deformed by non ideal optical systems in real time. In general the distortion of the wavefront from an ideal point source is measured, and a deformable mirror is used to compensate the aberrations and return the wavefront to a plane or spherical wave. This study concentrates on methods that omit wavefront sensing from the reconstruction process. Such methods use stochastic algorithms to find the extremum of a certain sharpness function, thereby correcting the image without any information on the wavefront. Theoretical work has shown that the optical problem can be mapped onto a model for crystal roughening which was the motivation for using the simulated annealing algorithm. The simulated annealing algorithm was evaluated in comparison to the stochastic parallel gradient descent algorithm (SPGD) which is a ”fast” variation of the traditional steepest descent algorithm. The usual adaptive optics method utilizing the Hartmann-Shack (HS) wavefront sensor was also experimentally implemented and evaluated for comparison. Both search algorithms were evaluated numerically by use of a specially designed simulation and visualization program and by laboratory experiments. We present a first hardware realization of the simulation annealing algorithm in a specially designed adaptive optics system intended to image the retina of the human eye. The results in simulations and laboratory experiments demonstrate the ability of the algorithms to correct deformed images in a narrow time span. In addition, an advantage of the simulated annealing over the SPGD algorithm in reaching a better solution was observed. The convergence rate of both algorithms was then improved by use of the mirror eigenmodes as calculated using the HS sensor.