Phase extraction neural network from gamma-distorted fringe images for three-dimensional surface measurement with single-frequency fringe projection
Optics Express · 24.08.2026 · 34 (17) · 32205
DOI: 10.1364/oe.600445 ↗Kurz gefasst
Fachartikel in der Zeitschrift Optics Express, online erschienen am 18.08.2026. Verfasst von Lifei Ren und 3 weiteren Autor:innen. Die Arbeit nennt 73 Literaturangaben. Gefördert von: Research Grants Council of the Hong Kong Special Administrative Region, China, National Natural Science Foundation of China, National Key Research and Development Program of China.
Abstract (Originalsprache)
This paper presents an accurate 3D measurement method for binocular fringe projection profilometry using a phase extraction neural network. To decrease the measurement error induced by the gamma effect of the structured light projector, this paper designs, what we believe to be, a novel artificial neural network with an arctangent layer to construct the nonlinear relationship between wrapped phase and grey values of three deformed fringe images and extract phase value pixel-by-pixel from gamma-distorted three-step fringe images. To achieve phase unwrapping while minimizing the number of projected fringe patterns, a Markov random fields-based phase unwrapping method is proposed by exploiting the internal functional relationship between the disparity and phase value as well as the local smoothness of the depth map. The point clouds of the measured objects are reconstructed by the stereo-matching of phase maps from two views. The experiments show that the proposed method is able to realize accurate measurements with a maximum mean absolute distance of 17.1 μm, and a corresponding standard deviation of 21.1 μm, with respect to the point clouds measured by a coordinate measuring machine over the field of view of 50mm × 50 mm, which only requires three fringe patterns, although the gamma effect of projectors exists.
Förderung
- Research Grants Council of the Hong Kong Special Administrative Region, China
- National Natural Science Foundation of China
- National Key Research and Development Program of China
- Hong Kong Polytechnic University
- European Partnership on Metrology programme
Zitieren
APA
Ren, L., Cheung, C. F., Li, D., & Cao, Y. (2026). Phase extraction neural network from gamma-distorted fringe images for three-dimensional surface measurement with single-frequency fringe projection. Optics Express, 34(17), 32205. https://doi.org/10.1364/oe.600445
Vancouver
Ren L, Cheung CF, Li D, Cao Y. Phase extraction neural network from gamma-distorted fringe images for three-dimensional surface measurement with single-frequency fringe projection. Optics Express. 2026;34(17):32205. doi:10.1364/oe.600445
BibTeX
@article{Ren2026,
author = {Ren, Lifei and Cheung, Chi Fai and Li, Da and Cao, Yanlong},
title = {Phase extraction neural network from gamma-distorted fringe images for three-dimensional surface measurement with single-frequency fringe projection},
journal = {Optics Express},
year = {2026},
volume = {34},
number = {17},
pages = {32205},
doi = {10.1364/oe.600445}
}
EUVIMED