Speckle Noise Reduction in Michelson Digital\ud Holography Using Known or Unknown\ud Reference Linear Phases and Image Processing

摘要

Speckle is caused by the illumination of an optically rough surface by coherent\udlight [1] which can cause problems in digital holography. Speckle in a reconstruction\udreduces visibility of details and may impede important measurements of an\udobject. Speckle can be reduced by capturing multiple holograms and summing the\udreconstructed intensities [2-4]. Digital signal processing methods, which are applied\udafter capture, have also been proposed [5–8].\udWith respect to three-dimensional (3D) objects, we solve this problem by outof-\udplane rotating the reference mirror in discrete steps, capturing a hologram each\udtime, and introducing image processing to rectify the resulting shifted diffraction\udfields. Each rotation amount need not be known and so can be generated by a very\udfast unstable device. This reference beam rotation is different from [4], where the\udobject beam is rotated at different angles. Changing the angle of incidence of the\udreference beam changes the phase implicitly captured in the hologram, and results in a different speckle pattern in each reconstructed object. The positions of the\udobject and the camera remain the same so there is no change in perspective between\udholograms, allowing integration of an arbitrary number of them using the\udfull rotation range allowed by the space-bandwidth product of the camera. The\udintensities of numerical reconstructions are averaged after image processing registration\udsteps, which produces speckle-reduced reconstructions that closely match\udtheoretical predictions for statistically independent speckle fields.