Spatial phase unwrapping approach for single-frame 3D shape measurement based on deep learning

Abstract

To address the challenge of balancing accuracy and speed in traditional phase unwrapping algorithms, this paper proposes a deep-learning-based single-frame spatial phase unwrapping method. By leveraging extensive data learning, two neural networks are trained to directly acquire phase information and modulation from a single-frame fringe pattern. Then, through the integration of a modulation sorting phase unwrapping algorithm, we achieve high-precision 3D surface reconstruction from a single-frame fringe pattern, thereby enabling rapid object measurement. The experimental results demonstrate the remarkable accuracy of the proposed method in phase unwrapping, approaching the level achieved by the 12-step phase-shifting method. The integration of deep learning into phase unwrapping offers promising prospects for further developments in this area. This advancement holds significant implications for high-speed measurement in the manufacturing field.