Background

Digital Image and Volume Correlation (DIC and DVC) are non-contact measurement techniques that are used during mechanical testing for quantitative mapping of full-field displacements. The relatively high noise floor of DIC and DVC, which is exasperated when differentiated to obtain strain fields, often requires some form of filtering. Techniques such as median filters or least-squares fitting perform poorly over high displacement gradients, such as the strain localisation near a crack tip, discontinuities across crack flanks or large pores. As such, filtering does not always effectively remove outliers in the displacement field.

Objective

This work proposes a robust finite element-based filter that detects and replaces outliers in the displacement data using a finite element method-based approximation.

Methods

A method is formulated for surface (2D and Stereo DIC) and volumetric (DVC) measurements. Its validity is demonstrated using analytical and experimental displacement data around cracks, obtained from surface and full volume measurements.

Results

It is shown that the displacement data can be filtered in such a way that outliers are identified and replaced. Moreover, data can be smoothed whilst maintaining the nature of the underlying displacement field such as steep displacement gradients or discontinuities.

Conclusions

The method can be used as a post-processing tool for DIC and DVC data and will support the use of the finite element method as an experimental–numerical technique.

中文翻译：

---

基于鲁棒有限元的数字图像和体积相关位移数据滤波器

背景

数字图像和体积相关性（DIC和DVC）是非接触式测量技术，在机械测试过程中用于全场位移的定量映射。DIC和DVC的较高本底噪声（在区分以获得应变场时会激怒）通常需要某种形式的滤波。诸如中值滤波器或最小二乘拟合之类的技术在高位移梯度上的性能较差，例如在裂纹尖端附近的应变局部化，裂纹侧面或大孔的不连续性。这样，滤波并不总是有效地去除位移场中的异常值。

客观的

这项工作提出了一个鲁棒的基于有限元的滤波器，该滤波器使用基于有限元方法的近似值来检测并替换位移数据中的离群值。

方法

制定了一种用于表面（二维和立体DIC）和体积（DVC）测量的方法。通过从表面和全体积测量获得的裂纹周围的分析和实验位移数据证明了其有效性。

结果

示出了可以以识别和替换离群值的方式对位移数据进行滤波。此外，可以在保持基础位移场的性质（例如陡峭的位移梯度或不连续性）的同时对数据进行平滑处理。

结论

该方法可以用作DIC和DVC数据的后处理工具，并支持将有限元方法用作实验数字技术。