Noise reduction in digital holography phase maps by phase-preserving discrete Fourier resampling

Several methods have been proposed to reduce the detrimental effects of coherent noise in holographic imaging. Among them, the use of spatial-frequency masking or resampling has been widely applied because of its low implementation complexity and well-studied trade-off between denoising effectiveness and spatial resolution. While the digital application of this method has been successfully demonstrated for intensity images, its application to phase maps fails. This work shows that the phase applicability of these methods depends on the use of resampling masks that strictly keep the zero-order spatial frequencies. Alternative masks are proposed that demonstrate effective single-shot noise reduction in experimental phase maps from digital holographic microscopy. The resulting method is potentially extendable to any other complex-valued-field retrieval technique.