GMDIC: a digital image correlation measurement method based on global matching for large deformation displacement...

Area of Science:

• Optical Measurement
• Mechanical Engineering
• Computer Vision

Background:

• Digital Image Correlation (DIC) is a non-contact optical method for full-field displacement and strain measurement.
• Traditional DIC faces limitations in large deformations, parameter setting, matching accuracy, and computational speed.
• Deep learning (DL) shows potential to overcome these challenges in DIC applications.

Purpose of the Study:

• To develop a new deep learning-based DIC method for accurate displacement field measurement in complex, large deformations.
• To enhance feature representation and address limitations of traditional DIC methods.

Main Methods:

• A novel deep learning network combining Swin-Transformer with multi-head attention and ECA attention module.
• Incorporation of positional information to boost feature representation capabilities.
• Development of a comprehensive displacement field dataset for training, simulating real-world conditions and complex deformations.

Main Results:

• The proposed DL-DIC model achieves displacement prediction accuracy comparable to traditional DIC in practical experiments.
• The model demonstrates superior performance over traditional DIC methods specifically in large displacement scenarios.
• The network effectively handles complex deformations and large-scale image data.

Conclusions:

• The proposed deep learning-based DIC method offers a robust solution for measuring displacement fields in complex large deformations.
• This approach significantly improves upon the accuracy and efficiency limitations of traditional DIC.
• The method holds promise for advanced applications in material science and engineering requiring precise deformation analysis.