研究目的
To investigate the root-mean square errors (RMSEs) of measured displacements in digital image correlation (DIC) due to matched or mismatched displacement mapping functions, and to examine the relationships between RMSE and subset size or image noise.
研究成果
The RMSEs of measured displacements decrease monotonically with increasing subset size for matched or overmatched shape functions but show an optimal subset size for undermatched functions where RMSE first decreases then increases. RMSE is directly proportional to image noise standard deviation and is influenced by speckle pattern and deformation state. A subset size of 61x61 to 101x101 pixels is recommended, and image pre-filtering is advised to mitigate noise effects. Future work should aim for intelligent subset size selection.
研究不足
The study focuses on displacement errors due to shape function mismatch and does not consider material properties (e.g., isotropic vs. anisotropic). The optimal subset size selection is manual, suggesting a need for intelligent automation in future work. The experiments are limited to specific deformation types and pattern sets, which may not cover all practical scenarios.
1：Experimental Design and Method Selection:
The study uses simulated and real experimental speckle patterns to investigate displacement measurement errors in DIC. The path-independent DIC technique with the first-order displacement mapping function is employed. Linear, quadratic, cubical, and high-order displacement deformations are applied to assess errors under matched and undermatched conditions. Gaussian noise with varying standard deviations is added to patterns to study noise effects.
2：Sample Selection and Data Sources:
Simulated speckle patterns are generated using computer methods (e.g., inverse mapping), and real patterns are obtained from experimental setups involving a drawing machine and CCD camera, as well as from the DIC Challenge dataset (Sample 14). Patterns have sizes such as 800x800 pixels and 2048x588 pixels, with specific displacement gradients imposed.
3：4). Patterns have sizes such as 800x800 pixels and 2048x588 pixels, with specific displacement gradients imposed.
List of Experimental Equipment and Materials:
3. List of Experimental Equipment and Materials: A drawing machine, CCD camera, and controller are used for real pattern acquisition. Computer software for pattern simulation and DIC analysis is implied but not specified. Speckle patterns are created using mark pens for real experiments.
4：Experimental Procedures and Operational Workflow:
Reference patterns are selected, and deformed patterns are generated by applying predefined displacement fields (e.g., linear with ux=6e-3, quadratic with uxx=6e-6). Gaussian noise is added with standard deviations of 1, 2, 3, or 4 Gy values. DIC is performed with varying subset sizes (e.g., 41x41 to 141x141 pixels) to compute displacements and calculate RMSEs. Multiple repeated experiments are conducted for statistical reliability.
5：6). Gaussian noise is added with standard deviations of 1, 2, 3, or 4 Gy values. DIC is performed with varying subset sizes (e.g., 41x41 to 141x141 pixels) to compute displacements and calculate RMSEs. Multiple repeated experiments are conducted for statistical reliability.
Data Analysis Methods:
5. Data Analysis Methods: Root-mean square error (RMSE) is used to quantify displacement errors. Relationships between RMSE, subset size, and noise standard deviation are analyzed graphically and tabulated. Conclusions are drawn based on trends observed in the data.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
