Speckle noise reduction in quantitative optical metrology techniques by application of the discrete wavelet transformation

摘要

Effective suppression of speckle noise content in interferometric data images can help in improving accuracy and resolution of the results obtained with interferometric optical metrology techniques. In this paper, novel speckle noise reduction algorithms based on the discrete wavelet transformation are presented. The algorithms proceed by: (a) estimating the noise level contained in the interferograms of interest, (b) selecting wavelet families, (c) applying the wavelet transformation using the selected families, (d) wavelet thresholding, and (e) applying the inverse wavelet transformation, producing denoised interferograms. The algorithms are applied to the different stages of the processing procedures utilized for generation of quantitative speckle correlation interferometry data of fiber-optic based optoelectronic holography (FOBOEH) techniques, allowing identification of optimal processing conditions. It is shown that wavelet algorithms are effective for speckle noise reduction while preserving image features otherwise faded with other algorithms.