Generic saturation-induced phase-error correction algorithm for phase-measuring profilometry

Intensity saturation causes partial incorrect intensities in captured images, leading to obvious phase errors in high-dynamic-range phase-measuring profilometry. Most existing methods require numerous projected patterns or additional hardware equipment to retrieve the three-dimensional shape. This p...

Full description

Saved in:
Bibliographic Details
Published inMeasurement science & technology Vol. 34; no. 9; p. 95006
Main Authors Wu, Zebo, Lv, Na, Tao, Wei, Zhao, Hui
Format Journal Article
LanguageEnglish
Published 01.09.2023
Online AccessGet full text
ISSN0957-0233
1361-6501
1361-6501
DOI10.1088/1361-6501/accf28

Cover

More Information
Summary:Intensity saturation causes partial incorrect intensities in captured images, leading to obvious phase errors in high-dynamic-range phase-measuring profilometry. Most existing methods require numerous projected patterns or additional hardware equipment to retrieve the three-dimensional shape. This paper proposes a comprehensive saturation-induced phase-error correction method by combining an average-phase compensation method, applying four-step phase-shifting (PS) patterns, with a phase repair method employing a total variation minimization (TVM) model. The periodic characteristic of the saturation-induced phase error is analyzed. The phase error can be efficiently compensated by averaging the initial and auxiliary phase, which is calculated utilizing a set of PS patterns with a phase offset of π /4. Furthermore, a judgment condition is provided to detect invalid points in overexposed shiny areas where the initial calculated phases are wrong. The corrected phases are repaired utilizing the TVM model from the compensated phase information surrounding invalid points. Simulations and experiments show that the proposed method can simultaneously correct the phase in non-uniform, high-reflectivity scenes and shiny areas with high accuracy using relatively few images. The phase error is reduced by nearly 80%.
ISSN:0957-0233
1361-6501
1361-6501
DOI:10.1088/1361-6501/accf28