Fast algorithms for phase-diversity-based blind deconvolution

Abstract

Phase diversity is a technique for obtaining estimates of both the object and the phase, by exploiting the simultaneous collection of two short-exposure optical images, one of which has been formed by further blurring regularized variant of the Gauss-Newton optimization method for phase diversity-based estimated when a Gaussian likelihood fit-to-data criterion is applied. Simulation studies are provided to demonstrate that the method is remarkably robust and numerically efficient.