Evaluation of denoising algorithms applied to the reduction of speckle in digital holography

摘要

In this article we compare image denoising algorithms applied to laser digital holography technique. The presented work focuses on reducing speckle noise contribution. The evaluation methodology lies on images of synthesized phases of which one controls the level and noise type. We retain five algorithms known for their efficiency in the field of image processing. These are: algorithms used for SAR filtering (Synthetic Aperture Radar), algorithms based on wavelets, the NLmeans algorithm recently proposed, the Wiener filter and the median filter. Three evaluation criteria are used to compare selected algorithms: the gain of SNR and quality index; we propose also a new one: the reconstructed phase error which is particularly relevant in the domain of digital holography.