What is Digital Image Correlation?

Overview

Digital image correlation (DIC) is a surface displacement measurement technique that can capture the shape, motion, and deformation of solid objects. Rudimentary DIC results are easy to obtain, but reliable, high-quality DIC results can be difficult to achieve.

The goal of digitalimagecorrelation.org is to equip new DIC practitioners with the basics of DIC so they can produce optimum measurements in a time-efficient manner. As a follow-on to this brief website, readers are strongly encouraged to continue their study of DIC with the “Good Practices Guide for Digital Image Correlation” from the International DIC Society.

The content of digitalimagecorrelation.org is organized for DIC learners to gather the most important information by reading from start to finish, while more experienced users can jump ahead to a section of interest. The subject matter is intentionally simplified to be accessible to engineers of all skill levels, including undergraduates. Mathematical definitions and derivations are avoided, but should be reviewed and understood by DIC practitioners once their basic skill level is established. At the conclusion of each section, suggestions are provided for more detailed resources.

Section 1: DIC fundamentals

The basic operation of DIC is tracking a pattern (often called a speckle pattern) in a sequence of images. The process of a DIC experiment (illustrated on the source site) can be divided into three steps:

Obtain a pattern on the sample for tracking.
Capture images of the sample during motion/deformation.
Analyze the images to compute the sample surface’s displacements.

The first image in the sequence is defined as the reference image, or the baseline to which the other images are compared. DIC matches the pattern between the reference image and a deformed image, and then calculates the pattern’s displacements between the reference and deformed images.

A simplification of the DIC analysis is illustrated conceptually as follows:

The reference image has a recognizable pattern of dots that will be tracked.
A portion of the pattern, called a subset, is selected for tracking. The term facet is also used instead of subset, but the guide adopts the term subset.
The center of the subset (the red dot, which is not part of the speckle pattern) is the place in the reference image from which the displacement will be calculated.
After the material is deformed from the reference image’s initial position, the subset in the deformed image is matched to the subset from the reference image.
Once the subset is matched, DIC calculates the subset center’s relative displacement between the reference and deformed images. The displacement here is the (small) difference between the blue and red dots.

The basic operation is then extended to multiple subsets and DIC points: the same procedure is repeated with multiple equally-sized subsets in a grid, yielding multiple points with displacement information (DIC points).

The displacement at each DIC point is a vector. For two dimensions of displacement, the components can be written in a Cartesian coordinate system as the horizontal displacement (u) and vertical displacement (v). Three dimensions of displacements (u, v, and w) can also be measured with a more complicated type of DIC that uses triangulation (discussed in the section on main types of DIC).

DIC is commonly utilized to study the mechanical properties of solids. One of the most common experiments for solid materials is a uniaxial tension experiment. The goal is to quantify how much the material deforms when a force is applied. Methods include:

Strain gauges
Extensometers (mechanical, laser, or optical)
DIC

Strain gauges and extensometers provide a single measurement of strain or displacement, while DIC can provide many displacement measurements across the material. With displacements across the material, DIC can capture local deformations that arise from inhomogeneity, cracking, stress concentrations, plastic instabilities, phase transformations, and other localized phenomena.

Subset and step sizes

Two important dimensions in a DIC calculation are the subset size and the step size:

Subset size: width and height of the subset square in the reference image (in pixels).
Step size: distance between subset centers (in pixels).

Key considerations:

Each subset should contain at least three speckles (Sutton, Orteu, Schreier. doi:10.1007/978-0-387-78747-3).
Larger subsets give better pattern matching (more uniqueness) but lower spatial resolution and higher computation time.
Smaller subsets improve spatial resolution but may reduce matching robustness.
Step size has a strong effect on spatial resolution: smaller step sizes yield more DIC data points and higher spatial resolution.

Spatial and temporal resolution limits

Two important planning questions:

What is the smallest displacement that this experiment can reliably measure?
What is the image exposure time that should be used?

For (1), the smallest displacement measurement is limited by image quality. In general:

DIC algorithms can detect sub-pixel displacements on the order of 0.01 px (theoretical).
Practical noise floor is on the order of 0.10 px.

Sub-pixel resolution is enabled by interpolation (often bicubic spline interpolation). Experimental variables introduce error, so the noise floor typically limits the smallest reliable displacement.

For (2), because the practical noise floor is about 0.10 px, exposure time should limit sample motion during exposure to < 0.10 px (or < 0.01 px as a conservative choice). Exceeding this leads to motion blur and degraded accuracy.

Example:

Sample displacement: 1 micron/second
Imaging resolution: 10 microns/px
DIC noise floor: 0.10 px × (10 microns/px) = 1 micron

So exposure time should be less than 1 second (or less than 0.1 second for 0.01 px).

Strain calculation

Computing strains is a common post-processing step using DIC displacements. In general:

Spatial strains are computed from displacements via spatial derivatives that include filtering.
This filtering can blur highly localized deformations with sharp features or discontinuities.
Sharp features can be smoothed out and hide key phenomena like slip bands (Stinville, et al. doi:10.1007/s11340-015-0083-4).

Section 2: Types of DIC algorithms

By displacement dimensionality

One way to categorize DIC algorithms is by the dimensions of the calculated displacements:

2-D DIC (2D-DIC)
- Uses images from one camera.
- Measures two displacement components (u, v) in a plane.
- Assumes sample deformations are constrained to a plane parallel to the camera.
- Out-of-plane motion can be a large source of error (Sutton, et al. doi:10.1016/j.optlaseng.2008.05.005).
- Lens and optical distortions (e.g., barrel distortions) introduce error.
3-D DIC (3D-DIC)
- Uses images from more than one camera.
- Depth is measured through triangulation (u, v, w).
- As long as the sample remains in focus, out-of-plane deformations are measured rather than introducing error.
- Lens distortions are corrected through calibration.
- Calibration connects image length scales to physical length scales accurately.
- In contrast, 2-D DIC typically uses a simpler, less accurate pixel-to-physical conversion.

Important note: 3-D DIC measures displacements on the surface of a material, not within the volume.

Digital Volume Correlation (DVC)

DVC extends DIC from pixels to voxels (3D pixels):

Measures displacements within a solid volume.
Requires imaging systems that can see inside materials.
Example imaging modalities: X-ray tomography, confocal microscopy.

Local vs global DIC

A second categorization is by pattern matching technique:

Local DIC
- Pattern is separated into multiple subsets that are individually matched.
- Historically introduced first and is more common.
- Many principles in the guide focus on this approach.
Global DIC
- Pattern is matched in one go using a finite-element–based approach.
- Uses the full-field response with continuity constraints.

Most principles apply to both, but subset-related details apply specifically to local DIC.

Section 3: Speckle patterning

DIC tracks features on the sample surface that collectively form the speckle pattern. Sometimes the natural surface features suffice, but usually an artificial speckle pattern is required.

Optimum speckle patterns meet the following conditions:

Speckle density ~50%
- Too few or too many speckles leads to features that are too big or too small (Reu. doi:10.1111/ext.12110).
- Example patterns can be generated with the Speckle Generator software from Correlated Solutions, Inc.
Pattern stability in the testing environment
- E.g., in high-temperature experiments, pattern should not decay or darken.
Softened speckle edges
- Avoid sharp, distinct edges with the background.
- Reduces aliasing at pixel level (Reu. doi:10.1111/ext.12139).
Good grayscale contrast
- Reduces error (Sutton, Orteu, Schreier. doi:10.1007/978-0-387-78747-3).
- Histogram visualization: two peaks (dark and bright) with broad separation; ideally a bimodal Gaussian distribution.
Speckle size 3×3 to 7×7 pixels
- At least 3×3 pixels to avoid aliasing (Reu. doi:10.1111/ext.12111).
- Not much more than 7×7 pixels to maintain point density (Reu. doi:10.1111/ext.12110).
- Range refers to smallest and largest speckles, not averages (Reu. doi:10.1111/ext.12110).
- Example estimate:
  12 mm / 2048 px * (3 to 7 px per speckle) = 18 to 41 microns per speckle.
Pattern adheres and deforms with the sample without adding significant mechanical stress
- Ideal speckles deform with the sample.
- Rigid speckles that move with the sample but do not individually deform add relatively small errors (Barranger, et al. doi:10.1111/j.1475-1305.2011.00831.x).
Random speckle positions but uniform size.
Pattern covers the area of interest on the sample surface.

Speckle patterning methods

Paint

Popular due to compliance and ease of application.
High-quality patterns can be sprayed quickly.
Black and white paints recommended for contrast.
White background with black speckles is preferred (better contrast retention; LePage, Shaw, Daly. doi:10.1007/s40799-017-0192-3).
For large deformations or high strain rates, perform experiments within 24–48 hours of painting (paint hardens and becomes less deformable over time; Reu. doi:10.1111/ext.12147).
Airbrushes (e.g., Iwata CM-B) can produce speckles ~10–100 microns by varying pressure.
Spray paint cans produce larger speckles (~100–1000 microns).

Inks and dyes

Suitable for hyperelastic materials (elastomers, polymers, biomaterials) where paint is not stretchy enough.
Inks/dyes can permeate materials.
Application methods: stamping, masking, spraying, stenciling.
Example: biological soft tissue stained with methylene blue and airbrushed with white paint (Lionello, et al. doi:10.1016/j.jmbbm.2014.07.007).
Permanent markers are also used in practice.

Powder particles

Useful for moist or sticky materials (e.g., silicone rubbers) where powders adhere better than paint.
Dark speckles: graphite powder.
White basecoat: alumina, talc, magnesium oxide.
Enables very small speckles: with filters and compressed air, patterns <10 microns on smooth samples (Jonnalagadda, et al. doi:10.1007/s11340-008-9212-7).

Laser engraving

Engraving the sample surface with a laser cutter can produce patterns that survive high-temperature tests (Hu, et al. doi:10.1007/s40799-018-0256-z).

Nanoparticles

For very small speckles (~20–100 nm) for scanning electron microscopy DIC.
Use self-assembled nanoparticles (Kammers and Daly, doi:10.1007/s11340-013-9734-5).

Lithographed patterns

Another method for small speckles with high control at the microscale (Cannon, et al. doi:10.1007/978-3-319-51439-0_34).

Speckle pattern gallery (descriptions)

Examples (as described on the source):

Carbon fiber with gold nanoparticles (Kammers and Daly, doi:10.1007/s11340-013-9734-5)
- Speckles (~50 nm) are random in position and uniform in size.
- Occasional clumps detract slightly from correlation quality.
Airbrush-painted speckle pattern on metal
- Good randomness in positions.
- Non-uniform speckle sizes (some too large, some too small).
- Likely still correlates but with less spatial resolution than optimal.

Section 4: Image capturing

The key data collection step for DIC is image capture. General good practices in photography/microscopy apply, with extra considerations for DIC.

Image magnification

Magnification depends on sample length scales and phenomena of interest.
DIC algorithms are length-scale independent; physical length scale enters via magnification.
DIC has been used from meter-scale (e.g., collapse of Mount St. Helens; Walter. doi:10.1130/G32198.1) down to atomic scale (TEM; Wang, et al. doi:10.1115/1.4031332).
Most commonly, cameras capture images for DIC.

Building a DIC setup: cameras, lenses, and lights

Key selection criteria:

Camera sensor
- Low noise, high quantum efficiency, high dynamic range.
- Historically CCD > CMOS, but new CMOS (e.g., Sony Pregius) have comparable performance.
- Color is not required and may be problematic; black-and-white cameras are best.
- Machine vision cameras are often suitable.
Lenses
- Low distortion.
- Telecentric lenses are ideal (consistent magnification across depth/field).
- Use mid-range apertures (e.g., f/5.6, f/8, f/11 on an f/1.8–f/22 lens) to avoid distortion.
Lighting
- Evenly distributed over the area of interest.
- Intense enough for sufficient exposure but not so intense as to saturate pixels.
- Avoid excessive heat: halogen lamps are bright but hot; LEDs are cooler (though high-intensity LEDs can generate heat).
- Cross polarization: place polarizers on lights and lenses at orthogonal orientations to increase contrast, reduce error, and attenuate saturated pixels (LePage, Daly, Shaw. doi:10.1007/s11340-016-0129-2).
- Use a fan to gently blow air and reduce heat-wave distortions (Jones, Reu. doi:10.1007/s11340-017-0354-3).
Mounting and alignment
- Rigidly mount cameras and lenses on an optics table or high-quality tripod.
- Minimize vibrations; clamp/tie/tape camera cables.
- For multi-camera systems, ensure cameras view the same area. Small differences in epipolar line heights can make calibration difficult (Reu. doi:10.1111/ext.12083).
Focus
- Critical for DIC.
- The key region in the area of interest must be best-focused.

Best practices for cleaning cameras

To remove dust:

Point the camera at a uniform, diffuse, bright light (e.g., light panel with diffuser). Increase exposure until the image is mostly saturated and move the camera; dark spots that don’t move indicate dust.
Shine a flashlight on the lens or sensor cover from different angles to reveal specks of dust.

If air dusting is insufficient, use new, clean lens tissues moistened with lens cleaning solution.

DIC setup example

An example 3-D DIC system includes:

Two cameras with two lenses.
Two lights in stereo with the cameras.
A fiber optic light for back-lighting a glass calibration grid.
A tensile sample in a mechanical testing frame with a mechanical extensometer on its gauge section.
A fan blowing air between cameras and sample to minimize heat waves (Jones and Reu. doi:10.1007/s11340-017-0354-3).

Section 5: Calibration

For 2-D DIC:

Calibration is a length scale conversion from pixel space to physical units.
Requires a line of known length (e.g., the horizontal field width, HFW).
Small errors in line length lead to large displacement errors; a precision ruler or accurate reference is recommended.

For 3-D DIC:

Cameras must be calibrated with respect to each other.
A line is insufficient; calibration uses a calibration grid (plane of known dimensions).
Speckle and grid examples can be generated with Speckle Generator and Target Generator from Correlated Solutions, Inc.

General best practices for 3-D calibration:

Use a dial indicator arm on a two-axis stage with magnetic base to hold the grid (or a lower-cost “third-hand” equivalent).
Capture 25–50 calibration image pairs (more can help but with diminishing returns).
Incrementally move and rotate the grid within the field of view between captures.
Set lighting so the grid has good contrast just below saturation.
For glass grids, use backlighting with a diffuse LED panel or indirect backlighting (e.g., shining on a white matte poster board).
For HFW < ~25 mm, use a glass calibration grid with laser-etched marks.
For HFW > ~25 mm, printed grids on paper can suffice. Firmly affix the printed grid to a flat, rigid substrate (e.g., PMMA sheet).

Section 6: Validation and error evaluation

DIC practitioners should ask: “How accurate are these measurements?”

DIC results can vary widely depending on:

Setup
Speckle pattern
Correlation parameters
Other conditions

It can be difficult to isolate individual error sources, but overall error can be estimated using:

Comparison with a second measurement
- Compare DIC displacements with:
  - Rigid body translations from a trusted source (e.g., Vernier micrometer, precision linear stage).
  - LVDTs, lasers, or other displacement techniques.
- Compare strains/deformations with:
  - Extensometers (mechanical or laser).
  - Strain gauges (limited to small strains).
Noise floor measurement (Type A errors)
- Capture repeated static images (no motion/deformation between images).
- Compute mean/distribution of displacements.
- This distribution represents the noise floor (often between about 0.01 and 0.1 px; non-zero due to noise/error) (Reu, doi:10.1007/978-3-319-22446-6_24).

Section 7: Further suggestions and codes

Readers are strongly encouraged to continue their study of DIC with the “Good Practices Guide for Digital Image Correlation” from the International DIC Society.

Below is a non-exhaustive list of open-source DIC codes (in alphabetical order):

SUN-DIC
- Repository: https://github.com/gventer/SUN-DIC
RealPi2dDIC
- Publication: https://doi.org/10.1016/j.softx.2020.100645
NCorr
- Website: https://www.ncorr.com
µDIC
- Publication: https://doi.org/10.1016/j.softx.2019.100391
DICe
- Repository: https://github.com/dicengine/dice

Section 8: Source and credits

This knowledge base article is a structured mapping of the content from:

Website: https://digitalimagecorrelation.org/

All technical descriptions, examples, references, and recommendations are credited to the authors and contributors of that site and the works cited therein.