Processing of Digital Holograms with 2D Windowed Fourier Transform: Study, Performance Analysis and GPU Implementation - Archive ouverte HAL
Communication Dans Un Congrès Année : 2017

Processing of Digital Holograms with 2D Windowed Fourier Transform: Study, Performance Analysis and GPU Implementation

Résumé

The two-dimensional windowed Fourier transform constitutes the core of an algorithm considered today as the state of the art in digital holography with regard to the reduction of speckle noise. In this article we describe the implementation of this algorithm in a GPU environment in order to improve its performance in computing speed. The two-dimensional windowed Fourier transform relies on the computation of a representation of the phase maps using a two-dimensional Fourier transform spatially located using a gaussian window. The reduction of the noise is effected by a threshold applied on the components of the representation. We present the windowed Fourier transform noise processing algorithm and its performances compared to eleven other denoising algorithms. Then description of its implementation in GPU environment is discussed. We present a comparison of the computing time performance with different CPU and GPU programming environments, including Matlab, C and TensorFlow.
Fichier non déposé

Dates et versions

hal-02378291 , version 1 (25-11-2019)

Identifiants

  • HAL Id : hal-02378291 , version 1

Citer

Silvio Montrésor, Florent Carlier, Pascal Picart. Processing of Digital Holograms with 2D Windowed Fourier Transform: Study, Performance Analysis and GPU Implementation. IEEE SENSORS 2017, Oct 2017, Glasgow, United Kingdom. pp.1-3. ⟨hal-02378291⟩
24 Consultations
0 Téléchargements

Partager

More