Parallel optimization of tridimensional deformation measurement based on correlation function constraints of a...

Area of Science:

• Mechanical Engineering
• Optical Metrology
• Computational Science

Background:

• Speckle registration in full-field deformation measurement suffers from low accuracy and slow processing, especially for large specimens.
• Existing methods struggle with redundant computations and time-consuming searches for homologous points.

Purpose of the Study:

• To develop a parallel optimization method for 3D deformation measurement.
• To address the challenges of accuracy and speed in speckle registration for slender and large specimens.

Main Methods:

• A novel correlation function using multi-camera network constraints to restrict homologous point search to epipolar lines.
• Bundling multiple cameras to reduce matrix dimensions and enable parallel processing.
• Utilizing GPU parallel mechanisms (Compute Unified Device Architecture) for accelerated subpixel search and stereo matching.

Main Results:

• Achieved calculation speed-up ratios of 20.390x for speckle sequence and 17.873x for stereo registration.
• Ensured out-of-plane displacement measurement accuracy higher than 0.179 mm within a 2m x 2m x 3m spatial range.
• Demonstrated significantly improved speed and accuracy in 3D deformation measurement.

Conclusions:

• The proposed parallel optimization strategy effectively overcomes the limitations of traditional speckle registration.
• This approach offers crucial applications for rapid, stable, and accurate 3D deformation measurements of large-scale objects.