Research map / Point clouds / Point cloud classification
3D Measurement of Complex Textured Objects Based on Bidirectional Fringe Projection
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes ELPC-PCF, which models defocus-induced phase errors at reflectivity edges, corrects them with bidirectional (horizontal and vertical) fringe projection plus epipolar constraints, and fuses the resulting point clouds to improve 3D measurement of textured objects.
Problem
Structured-light / fringe projection profilometry accuracy degrades on objects with complex textures, especially at color boundaries, because camera defocus convolves reflectivity and biases phase.
Approach
An error model linking phase error to phase and grayscale gradients at texture edges; bidirectional fringe projection supplies an extra epipolar constraint that lets the correction value be solved directly without PSF calibration, followed by averaging of point clouds from the two directions.
Key results
- ELPC-PCF reduces MAE and RMSE by up to 65.74% and 67.75% on the calibration plate versus HFP baseline.
- Relative to TEM on the calibration plate, ELPC-PCF further reduces MAE by 27.29% and RMSE by 33.74%.
- On the standard sphere, ELPC-PCF reduces MAE by 64.08% and RMSE by 66.82% compared with HFP.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Apply ELPC-PCF instead of TEM | FPP · Calibration plate | MAE: 0.397 -> 0.221 | Helped |
| Apply ELPC-PCF instead of TEM | FPP · Calibration plate | RMSE: 0.712 -> 0.348 | Helped |
| Apply ELPC-PCF instead of TEM | FPP · Card holder | MAE: 0.332 -> 0.234 | Helped |
| Apply ELPC-PCF instead of TEM | FPP · Postcard | MAE: 0.169 -> 0.101 | Helped |
| Apply ELPC-PCF instead of TEM | FPP · Sphere | MAE: 0.281 -> 0.153 | Helped |
| Apply ELPC-PCF instead of TEM | FPP · Vase | MAE: 0.380 -> 0.259 | Helped |
| Apply ELPC-PCF instead of OEM | FPP · Cylinder | RMSE: 0.395 -> 0.260 | Helped |
| Add ELPC phase correction (ablation) | FPP · Calibration plate | MAE: 0.645 -> 0.240 | Helped |
| Add point-cloud fusion (ablation) | FPP · Calibration plate | MAE: 0.645 -> 0.317 | Helped |
Surprising findings
- Phase error is maximal when the offset of the imaging center lies in the low-reflectivity region, not at the ideal edge center, so surrounding high-reflectivity pixels should be prioritized for correction.
- Dual horizontal fringe projection gives poor correction, indicating error stems mainly from complex textures rather than random noise.
Limitations
- Assumes Gaussian PSF and small defocus range; simplifying assumption b/a = -1 for point cloud fusion.
- Error analysis for low-reflectivity regions is affected by random noise, degrading performance on dark textures (vase).
- Evaluated on a single hardware setup with a few objects.
Experimental setup
Benchmarks: calibration plate card holder postcard standard sphere vase cylinder
Metrics: MAE RMSE computation time
Baselines: KE (kernel estimation) IFT (inverse Fourier transform) OEM TEM HFP VFP DHFP
Builds on: Rao and Da two-dimensional error model (TEM) Hu et al. one-dimensional error model (OEM) Zhang's calibration method Phase-shifting with CGC
Closest papers by meaning
- Causes and Corrections for Bimodal Multi-Path Scanning With Structured Light (CVPR 2019 · similarity 0.83)
- Sparse Multi-Path Corrections in Fringe Projection Profilometry (CVPR 2021 · similarity 0.82)
- GUSLO: General and Unified Structured Light Optimization (AAAI 2026 · similarity 0.81)
- A Unified Spatial-Angular Structured Light for Single-View Acquisition of Shape and Reflectance (CVPR 2023 · similarity 0.81)
- Coherence As Texture - Passive Textureless 3D Reconstruction by Self-interference (CVPR 2024 · similarity 0.80)
- Evaluate Geometry of Radiance Fields with Low-Frequency Color Prior (AAAI 2024 · similarity 0.80)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.