Deep photometric learning (DPL)

Tanaporn Na Narong; Denis Sharoukhov; Tonislav Ivanov; Vadim Pinskiy; Matthew Putman

doi:10.1117/12.2555925

5 March 2020 Deep photometric learning (DPL)

Tanaporn Na Narong, Denis Sharoukhov, Tonislav Ivanov, Vadim Pinskiy, Matthew Putman

Proceedings Volume 11281, Oxide-based Materials and Devices XI; 1128110 (2020) https://doi.org/10.1117/12.2555925
Event: SPIE OPTO, 2020, San Francisco, California, United States

Abstract

Photometric stereo is a common technique for 3D reconstruction by calculating the surface normals of an object from different illumination angles. The technique is effective to estimate height profile of static objects with large features, but often fails for objects with smaller features or in flat environments with small depressions. We propose a method using deep learning to perform 3D reconstruction of small features. Our method handles sample noise, uneven illumination, and surface tilt. We demonstrate decreased noise susceptibility on synthetic data and promising performance on experimental datasets. This approach enables rapid inspection and reconstruction of complex surfaces without the need to use destructive or expensive analysis methods.

Conference Presentation

Citation Download Citation

Tanaporn Na Narong, Denis Sharoukhov, Tonislav Ivanov, Vadim Pinskiy, and Matthew Putman "Deep photometric learning (DPL)", Proc. SPIE 11281, Oxide-based Materials and Devices XI, 1128110 (5 March 2020); https://doi.org/10.1117/12.2555925

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