Deep DIC: Deep Learning-Based Digital Image Correlation for End-to-End Displacement and Strain Measurement

Ru Yang,Yang Li,Danielle Zeng,Ping Guo

DOI: https://doi.org/10.1016/j.jmatprotec.2021.117474

2022-01-07

Abstract:Digital image correlation (DIC) has become an industry standard to retrieve accurate displacement and strain measurement in tensile testing and other material characterization. Though traditional DIC offers a high precision estimation of deformation for general tensile testing cases, the prediction becomes unstable at large deformation or when the speckle patterns start to tear. In addition, traditional DIC requires a long computation time and often produces a low spatial resolution output affected by filtering and speckle pattern quality. To address these challenges, we propose a new deep learning-based DIC approach--Deep DIC, in which two convolutional neural networks, DisplacementNet and StrainNet, are designed to work together for end-to-end prediction of displacements and strains. DisplacementNet predicts the displacement field and adaptively tracks a region of interest. StrainNet predicts the strain field directly from the image input without relying on the displacement prediction, which significantly improves the strain prediction accuracy. A new dataset generation method is developed to synthesize a realistic and comprehensive dataset, including the generation of speckle patterns and the deformation of the speckle image with synthetic displacement fields. Though trained on synthetic datasets only, Deep DIC gives highly consistent and comparable predictions of displacement and strain with those obtained from commercial DIC software for real experiments, while it outperforms commercial software with very robust strain prediction even at large and localized deformation and varied pattern qualities. In addition, Deep DIC is capable of real-time prediction of deformation with a calculation time down to milliseconds.

Image and Video Processing,Materials Science,Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The problems that this paper attempts to solve mainly focus on several key challenges of Digital Image Correlation (DIC) technology in material testing:

1. **Prediction stability under large deformations**: Traditional DIC methods become unstable in prediction results when dealing with large deformations or when the speckle pattern starts to tear. This limits their use in applications that require measurement of large strains, such as in super - stretchable materials.

2. **Long computation time**: Traditional DIC methods require a long computation time, which affects their ability for real - time applications.

3. **Low spatial resolution**: The output of traditional DIC methods is affected by filtering and the quality of the speckle pattern, resulting in low spatial resolution.

4. **Accuracy of strain - field prediction**: Traditional DIC methods calculate the strain field through the spatial derivative of the displacement field. In the case of high - spatial - frequency noise amplification, it is difficult to obtain high - precision strain - field prediction with this method.

To address these challenges, the paper proposes a new deep - learning - based method - Deep DIC. This method realizes end - to - end prediction of displacement fields and strain fields from image pairs through two Convolutional Neural Networks (CNNs), namely DisplacementNet and StrainNet. Specifically:

- **DisplacementNet**: It is used to predict the displacement field and adaptively track the Region of Interest (ROI).
- **StrainNet**: It directly predicts the strain field from the image input without relying on the prediction of the displacement field, thereby significantly improving the accuracy of strain - field prediction.

In addition, the paper also develops a new data - set - generation method for synthesizing realistic and comprehensive data sets, including the generation of speckle patterns and the deformation of speckle images. The synthetic data set generated by this method can not only train the model but also ensure the performance of the model on real experimental data.

In conclusion, the main goal of the paper is to achieve robust and high - precision prediction of displacement fields and strain fields under large deformations and different speckle - pattern qualities through the Deep DIC method, and significantly reduce the computation time to support real - time measurement and prediction.