研究目的
To provide simplified formulas for estimating the 3D reconstruction error in 3D digital image correlation (3D-DIC) systems based on known parameters without performing full DIC calculations, aiming to facilitate quick error assessment in practical applications.
研究成果
The study successfully derived simplified formulas for estimating 3D reconstruction error in 3D-DIC systems, which are based on key parameters like focal length, working distance, pixel size, stereo angle, and matching error. Numerical simulations and experiments validated that these formulas provide accurate error estimates, particularly for the center point, and can be used to guide system setup and parameter selection for desired measurement accuracy. The method offers a convenient way for a priori error estimation without full DIC computations, though it requires careful consideration of assumptions and limitations in practical applications.
研究不足
The theoretical derivation assumes perfect calibration, negligible lens distortion, and omission of high-order terms in pixel coordinates, which may not hold in all practical scenarios. The matching error is assumed to be Gaussian with mean 0 and std dev 0.01-0.02 pixels, but actual errors can vary. The formulas are primarily applicable to the center point of the field of view and may not fully capture errors in full-field measurements, especially at small stereo angles. Calibration accuracy and environmental factors are not fully accounted for in the error estimation.
1：Experimental Design and Method Selection:
The study involved numerical simulations to analyze 3D reconstruction error variations with stereo angle and measurement area, theoretical derivation to simplify error estimation formulas, and static experiments to validate the formulas. The pinhole camera model and stereo vision principles were employed.
2：Sample Selection and Data Sources:
A planar object with digital speckle patterns displayed on a tablet was used as the specimen. Calibration was done using chessboard images on the same tablet. Data included simulated and experimental pixel coordinates with added Gaussian noise.
3：List of Experimental Equipment and Materials:
Two Ueye CP3370 cameras (resolution: 2048 × 2048, pixel size: 5.5 μm × 5.5 μm), two Kowa 25 mm lenses, a tablet for displaying images, and software for 2D-DIC and 3D-DIC analysis.
4：5 μm × 5 μm), two Kowa 25 mm lenses, a tablet for displaying images, and software for 2D-DIC and 3D-DIC analysis.
Experimental Procedures and Operational Workflow:
4. Experimental Procedures and Operational Workflow: For simulation, parameters were set (focal length 25 mm, working distance 400 mm, etc.), Gaussian noise (mean 0, std dev 0.02 pixels) was added to pixel coordinates, and 3D positions were reconstructed and compared to known coordinates. For experiments, stereo angles from 10° to 120° were used, calibration images (100-150 pairs per angle) and speckle patterns were captured, matching error was calculated using 2D-DIC, and 3D reconstruction error was estimated using derived formulas and compared to 3D-DIC software results.
5：02 pixels) was added to pixel coordinates, and 3D positions were reconstructed and compared to known coordinates. For experiments, stereo angles from 10° to 120° were used, calibration images (100-150 pairs per angle) and speckle patterns were captured, matching error was calculated using 2D-DIC, and 3D reconstruction error was estimated using derived formulas and compared to 3D-DIC software results.
Data Analysis Methods:
5. Data Analysis Methods: Statistical analysis included mean and variance calculations of errors, comparison between theoretical and experimental results, and validation of error estimation formulas using least squares methods and Gaussian distribution assumptions.
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
