Which Pattern for a Low Pattern-Induced Bias?

The first objective of this presentation is to show that it is possible to explain the cause of the pattern-induced bias (PIB) observed in displacement fields obtained by local DIC. A model is presented for this purpose. It gathers the different errors made when retrieving this displacement by minimizing the optical residual over subsets. It is shown that PIB predicted with this model and its counterpart observed with displacement fields obtained with DIC are in good agreement. When DIC is applied on periodic patterns like checkerboards instead of random speckles, it is observed that PIB becomes negligible. Such regular patterns are however not well suited for DIC. Hence it is recalled how to process such images by minimizing the optical residual in the Fourier domain instead of the spatial one. PIB is assessed in this case, and it is also observed that PIB is negligible in displacement maps obtained with such regular patterns processed by minimizing the optical residual in the Fourier domain.

Automated summary

The first objective of this presentation is to explain the cause of the pattern-induced bias observed in displacement fields obtained by local DIC. A model is presented to gather the different errors in retrieving displacement. The study shows that using periodic patterns can significantly reduce the bias.