Ultra-low-latency 3D measurement with a 1D approximate de Bruijn pattern

Area of Science:

• Computer Vision
• 3D Imaging Technology
• Pattern Recognition

Background:

• Traditional 3D imaging systems often face trade-offs between speed and precision.
• Achieving low latency is crucial for capturing dynamic scenes and deformable objects.

Purpose of the Study:

• To introduce an ultra-low-latency projector-camera system for high-speed 3D imaging.
• To leverage the properties of one-dimensional de Bruijn patterns for enhanced performance.

Main Methods:

• Utilized a one-dimensional de Bruijn pattern, specifically four 5-bit sequences forming a 7-bit approximate sequence.
• Implemented GPU acceleration, taking advantage of the pattern's one-dimensional nature.
• Tested the system on diverse moving and deformable objects with varying material properties.

Main Results:

• The system demonstrated excellent performance across various object types and conditions.
• Achieved an average precision of 0.80 mm and an ultra-low latency of 0.059 ms for planar surfaces.
• Acquired 120.0K points per image on planar surfaces, balancing code length and decoding speed.

Conclusions:

• The proposed de Bruijn pattern-based system offers a significant advancement in 3D imaging speed and accuracy.
• The system effectively balances precision and decoding speed, making it suitable for dynamic 3D reconstruction.
• The GPU-accelerated, low-latency approach is highly effective for capturing complex, moving objects.