Related Experiment Video
Updated: Jul 1, 2026
11:34
High-resolution, High-speed, Three-dimensional Video Imaging with Digital Fringe Projection Techniques
Published on: December 3, 2013
16.3K
MSPANet: a multi-scale phase-aware network for spatial phase unwrapping via fringe order classification
Optics Express
|May 4, 2026
Summary
This study introduces a new network for 3D reconstruction, improving spatial phase unwrapping. The method enhances accuracy in complex scenarios by treating fringe-order estimation as a segmentation task.
Area of Science:
- Computer Vision
- 3D Reconstruction
- Computational Imaging
Background:
- Structured-light 3D reconstruction relies on accurate spatial phase unwrapping.
- Traditional methods struggle with noise and discontinuities, leading to errors.
Purpose of the Study:
- To develop a robust and accurate phase unwrapping method for 3D reconstruction.
- To address limitations of regression-based approaches in fringe-order estimation.
Main Methods:
- Introduced a Multi-Scale Phase-Aware Network (MSPANet).
- Formulated fringe-order estimation as a semantic segmentation problem.
- Integrated multi-scale feature extraction, dilated convolution, edge enhancement, and a phase-aware gating module.
Main Results:
- MSPANet demonstrated robust and accurate phase unwrapping on simulated and real-world data.
- The network effectively handled noise and discontinuities inherent in 3D reconstruction.
- Achieved reliable 3D shape measurement in complex environments.
Conclusions:
- MSPANet offers a significant advancement in spatial phase unwrapping for 3D reconstruction.
- The semantic segmentation approach overcomes limitations of prior methods.
- Enables more reliable and precise 3D shape measurement in challenging conditions.
