Object reconstruction from noisy holograms

The inappropriate processing and digitization of an optical hologram to become a digital hologram can be modeled by additive noise superimposed on the hologram and followed by a clipping operator. Direct reconstruction of this noisy digital hologram will usually yield unsatisfactory results because the noise will be the dominating term on the reconstruction plane. The constrained iterative twin image elimination algorithm for an in-line Fresnel hologram is modified by incorporating the noise constraint into the hologram constraint domain to estimate the additive noise function while eliminating the twin image. The result of the object reconstruction is shown to be improved.