digital-image-correlation

name	digital-image-correlation
description	Digital Image Correlation (DIC) — subset correlation, displacement field, full-field strain measurement, speckle pattern, camera calibration, stereo-DIC, error sources, ASTM E2208, structural testing applications.
metadata	{"priority":7,"promptSignals":{"phrases":["digital image correlation","DIC","full field strain","speckle pattern","displacement field measurement","stereo DIC"],"minScore":3}}

Digital Image Correlation (DIC) — Complete Skill

DIC Fundamentals

DIC: optical full-field measurement technique; tracks surface pattern (speckle) deformation between reference and deformed images → displacement and strain fields

Process:

1. Apply random speckle pattern to specimen surface
2. Capture reference image (unloaded)
3. Apply load/deformation
4. Capture deformed image(s)
5. Correlate subsets → displacement field u(x,y), v(x,y)
6. Differentiate displacement field → strain field ε_x, ε_y, γ_xy

Advantages over strain gauges: Full-field (thousands of data points vs. point measurements); non-contact; no specimen modification (except speckle); reveals strain concentrations, gradients

Correlation Algorithm

Subset (window): small region of pixels (typically 25×25 to 71×71 pixels) in reference image Correlation criterion (cross-correlation, normalized): C_NCC = Σ [f(x,y) × g(x',y')] / √(Σ f² × Σ g²) [maximized when subset matches] Or: Sum of Squared Differences (SSD) criterion (minimized): C_SSD = Σ [f(x,y) - g(x+u,y+v)]²

Shape functions: Affine: 6 DOF per subset (translation + rotation + shear) Higher-order: 12 DOF (includes quadratic terms for large deformation gradients)

Sub-pixel accuracy: Newton-Raphson iteration to find exact u, v to sub-pixel precision Typical accuracy: 0.01–0.05 pixels displacement; 50–200 μstrain for strain

Speckle Pattern Requirements

Good speckle properties:

• Random (no periodic pattern → aliasing)
• Isotropic (patterns visible in all directions)
• High contrast (dark on light background; or inverse)
• Feature size: 3–5 pixels/speckle (too small → aliased; too large → poor correlation)
• Coverage: 40–60% dark area

Speckle application methods: White base coat (matte) → black spray paint (fine droplets from distance 30–60 cm) Airbrush: finest control; 0.01–5 mm speckle size Laser speckle: naturally random from laser illumination; no preparation needed; small scale only

Scale calibration: Physical size / pixel count → mm/pixel; calibrate with ruler in field of view Sub-millimeter: microscope objective; SEM imaging for nano-scale DIC

Camera Setup

2D-DIC (Single Camera)

Assumptions: planar specimen; motion in plane only; no out-of-plane Errors from out-of-plane motion: ε_error ≈ Δz / z₀ [Δz = out-of-plane displacement; z₀ = working distance] For Δz/z₀ < 0.005 → error < 500 μstrain (often acceptable)

Camera requirements: Resolution: 5–20 Mpixel for good spatial resolution Frame rate: match test rate; quasi-static: 1 Hz; dynamic: 100–10,000 Hz Lens: telecentric lens (for best accuracy; eliminates perspective distortion); or calibrate and correct

Stereo-DIC (3D-DIC, Two Cameras)

Configuration: two cameras at 15–35° stereo angle Calibration: ZhangCalib or Bouguet toolbox; calibration plate with dots/circles Output: 3D displacement field (u, v, w); eliminates out-of-plane errors Camera calibration parameters: intrinsic (focal length, principal point, distortion) + extrinsic (rotation, translation between cameras)

Calibration target requirements: Calibration plate accuracy: < 1/10 of measurement accuracy Standard: ASTM E2208 recommends verified calibration targets

Strain Computation

Strain from displacement gradient (Green-Lagrange for large deformation): ε_xx = ∂u/∂x + 0.5 × [(∂u/∂x)² + (∂v/∂x)²] [finite strain]

Engineering strain (small deformation): ε_xx = ∂u/∂x; ε_yy = ∂v/∂y; γ_xy = ∂u/∂y + ∂v/∂x

Numerical differentiation: Central difference: ε_x ≈ [u(x+Δ,y) - u(x-Δ,y)] / (2Δ) [noise sensitive] Polynomial fit: fit polynomial to displacement field in region → differentiate analytically (smoother)

Virtual extensometer: Average strain between two points = Δu / L₀ [equivalent to physical extensometer; select any L₀] Higher accuracy than local subset strain for uniform fields

Strain accuracy: Subset size 51×51 px: ε accuracy ≈ 100 μstrain at 5 Mpixel camera Subset size 25×25 px: ε accuracy ≈ 200–400 μstrain (smaller subset → more spatial resolution but more noise)

Error Sources

Systematic errors: Lens distortion: corrected by calibration; residual < 0.05 px Lighting variation: changes correlation function → apparent displacement; use LED ring light (stable) Temperature: thermal expansion of camera, lens, specimen → spurious strain; thermal shielding Subset size too small: poor correlation → noise; too large: misses strain gradient

Random errors: Camera noise (shot noise, read noise): 0.01–0.05 px for modern cameras Quantization (8-bit vs. 16-bit): 16-bit → much better displacement precision

Strain error floor: depends on displacement noise σ_u and subset size S: σ_ε ≈ σ_u / (0.5 × S × pixel_size) [reduces with larger subset; increases with noise]

Applications

Fracture mechanics (K-field mapping): J-integral from DIC strain field: J = ∫ (W n₁ - T_i ∂u_i/∂x₁) dΓ Stress intensity K from fitting Williams expansion to displacement field

Tensile testing correlation: ASTM E8/E83 accepts DIC for non-contact extensometry (virtual extensometer) Required for brittle materials (glass, ceramics) where contact clips fail

Fatigue crack monitoring: Track crack tip opening displacement (CTOD) per cycle → J vs. N → Paris law Spatial resolution: < 10 μm achievable with macro-lens or microscope

Digital volume correlation (DVC): 3D extension using CT images; voxel correlation; measures internal strain fields in foam, composites

DIC Software

Commercial: VIC-2D, VIC-3D (Correlated Solutions); ARAMIS (GOM/Zeiss); StrainMaster (LaVision) Open source: pyDIC, NCorrPy, OpenDIC, Ncorr (MATLAB)

Post-processing: export strain maps as contour plots; extract virtual extensometers; compare with FEA

Standards

Standard	Scope
ASTM E2208	Standard guide for evaluating non-contacting optical strain measurement systems
ASTM E8/E8M	Tension testing with DIC extensometry
ISO 6507	Hardness testing (DIC applied near indent)
DIC Challenge	International round-robin for DIC accuracy benchmarking (iDICs)

Output

Provide: DIC type (2D/stereo-3D), camera resolution [Mpixel] and frame rate [fps], working distance [mm] and field of view [mm×mm], spatial resolution [mm/pixel], speckle size [pixels] and method, subset size and step [pixels], displacement accuracy [pixel] and [μm], strain accuracy [μstrain], calibration method (ZhangCalib, plate used), out-of-plane error assessment (if 2D), virtual extensometer gauge length [mm] and strain [%], comparison with strain gauge (if applicable), peak strain field location, fracture parameters extracted (K, J, CTOD) if applicable, and applicable standard (ASTM E2208, E8/E8M).