Deep learning-based speckle decorrelation denoising for wide-field optical metrology

Silvio Montrésor; Marie Tahon; Antoine Laurent; Pascal Picart

doi:10.1117/12.2556545

1 April 2020 Deep learning-based speckle decorrelation denoising for wide-field optical metrology

Silvio Montrésor, Marie Tahon, Antoine Laurent, Pascal Picart

Proceedings Volume 11352, Optics and Photonics for Advanced Dimensional Metrology; 113520R (2020) https://doi.org/10.1117/12.2556545
Event: SPIE Photonics Europe, 2020, Online Only

Abstract

This paper presents a deep-learning-based algorithm dedicated to the processing of speckle noise in phase measurement from holography. The deep learning architecture consists in a pre-trained residual convolution neural network, initially devoted to de-noising of natural images. In order to adapt the network to de-noise phase maps, a database is constituted by a set of noise-free and noisy-phase maps corresponding to realistic noise conditions (non-Gaussian, non-stationary, controlled speckle size). The algorithm is qualified according to quantitative metrics such as the phase error, the error method, the Qindex and the computation time. This paper demonstrates that the proposed de-noising algorithm yields stateof-the-art results in terms of phase error and error method. In addition, the processing efficiency in term of computation time appeared to be better. So, de-noising of phase maps using such deep-learning-based approach is expected to yield very promising results in optical metrology. Application of the method to the characterization of vibrations over surface about 400cm² is presented when dealing with vibration amplitude of 20nm at 17kHz recorded at 100kHz by a high speed in-line Fresnel configuration.

Conference Presentation

Citation Download Citation

Silvio Montrésor, Marie Tahon, Antoine Laurent, and Pascal Picart "Deep learning-based speckle decorrelation denoising for wide-field optical metrology", Proc. SPIE 11352, Optics and Photonics for Advanced Dimensional Metrology, 113520R (1 April 2020); https://doi.org/10.1117/12.2556545

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available