Fourier-based interpolation bias prediction in digital image correlation.

Area of Science:

• Digital Image Correlation
• Optical Metrology
• Image Processing

Background:

• Interpolation bias is a known issue in digital image correlation (DIC).
• Existing methods lack a clear understanding of the underlying causes of this bias.
• Quantifying and mitigating interpolation bias is crucial for accurate strain measurement.

Purpose of the Study:

• To derive analytic formulae for interpolation bias in DIC using the Fourier method.
• To introduce the concept of the interpolation bias kernel for characterizing frequency response.
• To develop a predictive approach for interpolation bias.

Main Methods:

• Fourier method for deducing analytic formulae of interpolation bias.
• Introduction of the interpolation bias kernel to analyze frequency response.
• Development of a prediction approach using speckle spectrum and interpolation transfer function.
• Novel experimental subpixel translation technique.

Main Results:

• Sinusoidal interpolation bias curves are explained by aliasing effects.
• High-frequency components are identified as the primary source of interpolation bias.
• A predictive approach achieves significant acceleration and agrees with simulations and experiments.
• Experimental results show accuracy up to 0.01 pixels.

Conclusions:

• The interpolation bias kernel provides a measure of subset matching quality.
• The proposed prediction method is effective and efficient.
• The novel experimental technique eliminates mechanical errors for high-accuracy measurements.