High-dynamic-range fringe projection profilometry using a double-exposure branch fusion network

Area of Science:

• Optics and Photonics
• Computer Vision
• Metrology

Background:

• Manufacturing requires precise 3D measurement of high-reflective workpieces.
• Low dynamic range in images causes fringe information loss, hindering accurate 3D measurement.
• Existing 3D measurement techniques are often slow, imprecise, and lack dynamic range.

Purpose of the Study:

• To develop a high-dynamic-range (HDR) fringe projection profilometry method for accurate 3D measurement of high-reflective workpieces.
• To address the challenges of fringe information loss and point-cloud loss in reflective surface measurement.
• To improve the precision, efficiency, and dynamic range of 3D measurement systems.

Main Methods:

• A double-exposure branch fusion network (DBF-Net) was proposed for HDR fringe projection profilometry.
• A sinusoidality-preserving algorithm was embedded within DBF-Net to generate HDR fringe images.
• A weighted fusion loss function was designed to enhance network generalization.
• Camera-projector inverse calibration established an accurate object-image relationship.

Main Results:

• The proposed DBF-Net achieved a measurement accuracy of 0.06 mm for high-reflective workpieces.
• The method effectively resolved point-cloud loss issues common with reflective surfaces.
• Comparative experiments demonstrated superior image quality, phase demodulation accuracy, and 3D reconstruction compared to existing methods.

Conclusions:

• DBF-Net offers a high-precision and efficient solution for 3D measurement of challenging reflective surfaces.
• The developed method significantly advances fringe projection profilometry capabilities.
• This technique is crucial for quality control and inspection in advanced manufacturing industries.