Notice: the most up-to-date and maintained version of FIDIC is qDIC!
The Fast Iterative Digital Image Correlation Algorithm (FIDIC) is a 2D version of FIDVC algorithm (please see Bar-Kochba, Toyjanova et al., Exp. Mechanics, 2014 for more details) to find dispalcements fields in a 2D image.
The following implementation contains the MATLAB m-files for our FIDIC algorithm along with synthetic example images. The FIDIC algorithm determines the 2D displacement fields between consecutive images or from a static reference image to a current image.
MATLAB 2011b (for "griddedInterpolant") and the associated Image Processing Toolbox (for other miscellaneous function calls) are the minimum supported requirements to run this code. Under some circimstances older versions may function using "interpn", but performance and/or accuracy may suffer. Our development is currently under Matlab 2015b on CentOS 7 and Window 7 x64.
- To check if the images have the required speckle pattern and intensity values for correlation please use our DIC simulator.
- We recommend that the input image stack should have at least 3 times the subset size as the number of pixels in each dimension. The default subset size is 64x64, meaning the the minimum input image size should be 192x192.
- Non-square images are acceptable
- The fundamental image type used for input is .mat
- Out-of-the-box FIDIC supports TIF images with
img2mat.m, other file formats require simple modification
- Make sure that the main files and the supplemental m files (from file exchange) are in the current (working) directory for MATLAB.
- Copy the desired test images from the
stressortranslationsubdirectories to thetest_imagesdirectory. - Run the
exampleRunFile.mfile to get 2D displacement fields between the two images. Note that the displacement output is in the form of either a three pixel translation or a generic uniaxial tension test, depending on the test image set selected.
- We recommend that the input image size in each dimension be at least three times the size of the subset size. The default subset size is 64x64, so we recommend that the minimum input image size should be 192x192.
- Make sure that the main files and the supplemental m-files (from file exchange) are in the working directory on Matlab.
- Run the exampleRunFile.m file to and compare its displacement outputs to the contour plots expectation for each image type.
- Function files
- addDisplacements_2D.m
- checkConvergenceSSD_2D.m
- DIC.m
- filterDisplacements_2D.m
- funIDIC.m
- IDIC.m
- removeOutliers_2D.m
- areaMapping_2D.m
- Supplemental .m files from the MATLAB file exchange:
- inpaint_nans.m
- mirt2D_mexinterp.m (Optional, not currently in use)
- Example files to run basic DIC
- exampleRunFile.m
- img2mat.m
- imageCropping.m
- FIDIC_plot.m
- image_eval.m
- Example test images
What are the requirements for the input images?
Please refer to input image requirement.
Can I use FIDIC for finding displacement fields in 3D images?
No. But you can use FIDVC, this finds 3D displacements in 3D image stack.
Why does the example fail to run?
In many cases where the example images fail to run, the minium specifications for MATLAB have not been met.
@article{bar2014fast,
title={A fast iterative digital volume correlation algorithm for large deformations},
author={Bar-Kochba, E and Toyjanova, J and Andrews, E and Kim, K-S and Franck, C},
journal={Experimental Mechanics},
pages={1--14},
year={2014},
publisher={Springer}
}For questions, please first refer to FAQ and Questions/Issues. Add a new question if similar issue hasn't been reported. We shall help you at the earliest. The author's contact information can be found at Franck Lab.