Phase Retrieval with Roughly Known Mask

Uniqueness, up to a global phase, is proved for phasing with a random roughly known mask (RKM) of high uncertainty. Phasing algorithms alternating between the object update and the mask update are systematically tested and demonstrated to have the capability of nearly perfect recovering the object and the mask (within the object support) simultaneously for up to 50% mask uncertainty. The uncertainty threshold for successful recovery can be further lifted to more than 70% under the additional sector condition on the object. The phasing algorithms with RKM are robust to low resolution masking as well as other types of noise. © 2012 Optical Society of America