Digital Image Correlation, abbreviated as “DIC”, is an optical, “measurement technique” used to determine the full-field, shape (contour), displacement and strain for experimental solid mechanic applications in materials testing.
Point Tracking, abbreviated as “PT”, is an optical, “generalized measurement approach” used to determine discrete kinematic quantities of points, such as; rotations, velocities & accelerations, for motion and/or vibration analysis.
Some advantages of DIC & Point Tracking compared to other measurement techniques (ie. strain gauges) include:
- DIC measures an entire (optically) visible image yielding full-field (continuous) results of shape (contour), displacement and strain à allowing more quantitative, large-scale information to be acquired, as opposed to measuring point-based results of single mechanical properties such as; one-axis directional displacements (ie. from LVDTs) or major strains (ie. from Strain Gauges).
- DIC is a non-contact method that requires no mechanical connection or physical interaction with the test object surface à allowing measurement of delicate & fragile objects and (moist or wetted) biomaterials but also saves time in the preparation of adhering strain gauges to a test object surface.
- DIC is versatile as it has no measurement speed and/or no mechanical limitations à allowing anything that the camera “sees” in shape (flat, curved, cylindrical, spherical, bi-planar), or size (from mm² to m²), or motion (up to MHz) to be measured. Additionally, there are no fixed gauge lengths for the technique à one DIC system can be used for multiple, various applications.
- DIC resolves measurements within sub-pixel accuracy allowing micro strains to be determined both in-plane (parallel to a surface), and using a double-sided setup, out-of-plane (perpendicular to a surface).
Both Digital Image Correlation (DIC), as the name itself implies, and Point Tracking, work on the principle of image “correlation”, whereby the relationship between digitally acquired images taken over both time (temporal) and perspective space (spatial) is determined, using a recognition algorithm. Recognition algorithms (such as DIC and Marker Tracking) require “features” to be optically visible on the test object surface, which are “matched” + “tracked” (=“correlated”) across a measurement series (“steps”). Features are contrast textures of information that can be free-hand, random patterns or fiducial markers (aka. artificial artifacts).
The Measurement Procedure of DIC & Point Tracking
A “camera” is used to “digitize” the optically visible features on the test object surface, on each “pixel”, as “discrete grey-level value intensities”.
A “reference step” is acquired by the user, that defines the “reference-state” of the feature pattern” at a certain time-point. Usually, the reference state is typically selected when the test object is both stationary and unloaded. In one or more “system camera/s”, image “steps” are successively acquired in synchronization of the temporally deforming test object and saved in a “test measurement series”.
For correlation, the feature pattern reference-state MUST be optically visible throughout the test.
The measurement space is parameterized by a ”rectangular grid” of equidistant centre (data) points. In the simplest case, the grid is aligned to the image of one of the system cameras. This distinguished camara is then referred as the “reference camera”.
The spacing between the grid points (aka. the grid pitch) is adjustable by the “Grid Spacing/Step Size” [px].
Each facet/subset is tracked throughout the measurement series by using a “subset shape matching algorithm”, which is implemented to “search”, “find” and “define” the new position (coordinates) and shape (displacement) of the deformed facets/subsets in all measurement steps and in all system camera/s.
The subset shape matching algorithm aims to align a shape function between the reference and deformed facet/subset through identification of the minimized difference between the grey-level intensities.
This yields the 2D-coordinates & 2D-displacement information of the facet/subset for each system camera. This process is referred to as “temporal correlation”.
By using a “reference object” with known geometric qualities, called a “calibration target”, a process known as a “projection calibration” is performed by the user.
Calibrating the projection calibration involves the software building (and optimizing) a 3D-virtual space model, called a “projection calibration model”.
The model parametrically describes the “optical setup” of each system camera (“intrinsic” properties) and each system camera within the entire system (“extrinsic” properties).
Using the 2D-pixel locations of the “correlated grid points” with the projection calibration model, the intersection of the ray lines from all cameras of the system can be resolved as an object point with an absolute 3D-coordinate on all other system cameras. This process is known as the “triangulation”.
The 3D-displacements can be resolved through calculating the (relative) difference of the absolute 3D-coordinates over the measurement series. From the absolute 3D-coordinates and 3D-displacements, the “deformation gradient” of each grid point can be determined, and thus the surface strain can be retrieved.
