Study of the performance of different subpixel image correlation methods in 3D digital image correlation

Area of Science:

• Mechanical Engineering
• Materials Science
• Optical Metrology

Background:

• Three-dimensional digital image correlation (3D-DIC) is a widely used technique in engineering and manufacturing for non-contact strain measurement.
• Errors arising from different image matching algorithms in 3D-DIC are not extensively studied.
• Accurate 3D reconstruction relies on precise correlation algorithms and reliable speckle pattern simulation.

Purpose of the Study:

• To propose a 3D speckle image generation algorithm that minimizes interpolation bias.
• To investigate and compare the performance of various subpixel correlation algorithms for 3D-DIC.
• To establish an error criterion for 3D reconstruction and recommend optimal algorithms.

Main Methods:

• Development of a 3D speckle image generation algorithm without texture interpolation bias.
• Simulation of 3D speckle images before and after deformation.
• Performance evaluation of four different subpixel correlation algorithms using simulated data.
• Proposal of a 3D reconstruction error criterion.

Main Results:

• The proposed algorithm effectively omits interpolation errors in speckle simulation.
• Performance analysis revealed significant differences among the tested subpixel correlation algorithms.
• A quantitative error criterion for 3D reconstruction was established.
• The first-order Newton-Raphson iteration method and gradient-based method demonstrated superior performance.

Conclusions:

• The developed 3D speckle generation method enhances the reliability of 3D-DIC simulations.
• Algorithm selection is critical for minimizing errors in 3D-DIC measurements.
• Newton-Raphson and gradient-based methods are recommended for precise 3D-DIC applications.