Speckle denoising by variant nonlocal means methods

Yassine Tounsi, Manoj Kumar, Abdelkrim Nassim, Fernando Mendoza-Santoyo, Osamu Matoba

Research output: Contribution to journalArticlepeer-review

28 Scopus citations


This study aims to demonstrate the performances of nonlocal means (NLM) and their variant denoising methods, mainly focusing on NLM-shaped adaptive patches and several NLM-reprojection schemes for speckle noise reduction in amplitude and phase images of the digital coherent imaging systems. In the digital coherent imaging systems such as digital speckle pattern interferometry, digital holographic interferometry, etc., the image quality is severely degraded by additive uncorrelated speckle noise, due to the coherent nature of the light source, and therefore limits the development of several applications of these imaging systems in many fields. NLM and its variant denoising methods are employed to denoise the intensity/phase images obtained from these imaging systems, and their effectiveness is evaluated by considering various parameters. The performance comparison of these methods with other existing speckle denoising methods is also presented. The performance of these methods for speckle noise reduction is quantified on the basis of two criteria matrices, namely, the peak-to-signal noise ratio and the image quality index. Based on these criteria matrices, it is observed that these denoising methods have the ability to improve the intensity and phase images favorably in comparison to other speckle denoising techniques, and these methods are more effective and feasible in speckle-noise reduction.

Original languageEnglish
Pages (from-to)7110-7120
Number of pages11
JournalApplied Optics
Issue number26
StatePublished - 10 Sep 2019
Externally publishedYes

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Engineering (miscellaneous)
  • Electrical and Electronic Engineering


Dive into the research topics of 'Speckle denoising by variant nonlocal means methods'. Together they form a unique fingerprint.

Cite this