Hybrid algorithm in phase diversity wavefront sensing

dazun zhao,heng mao,xiao wang
DOI: https://doi.org/10.1117/12.831453
2009-01-01
Abstract:Phase diversity (PD), proposed by Gonsalves, is a kind of wavefront sensing technique based on measurement of two or more images of object. The optical system involved is relatively simple. It makes use of the methods of optimization and image processing, which can jointly estimate phase aberration as well as object itself simultaneously. The most significant characteristic of this technique is that it works well with extended scenes. Steepest descent method and conjugate gradient method both are preferable algorithms for nonlinear optimization. As a matter of fact, any one of two methods has some limitations. Steepest descent method is a local property and conjugate gradient method's convergence rate is slow. Combining two methods to develop a mixed algorithm, we can avoid entrapping into a local minimum and raise global convergence rate. Simulation results demonstrate that the hybrid algorithm has the features of quick convergence rate, comparatively large convergence range, which make the method of phase diversity remarkably robust and numerically efficient.
What problem does this paper attempt to address?