Pattern for Large Structures using DIC

What?

Digital Image Correlation is a widely used technique to measure the shape, displacement, and strains of a structure by using at least two cameras and triangulating the position of each point in the field-of-view.
In order to differentiate different parts of the tested sample, a pattern is applied to the structure, usually using black paint speckles on a white background.
For large structures, it is difficult to apply paint as speckles may be too small to detect. Differences in the size and density of the speckles may also create non-homogenous errors.
A technique using Perlin Noise can create a reliable, homogenous pattern with predictable feature sizes.

Perlin Noise is a simple gradient-based noise used in CGI to create clouds, terrains, or other textures.
It is based on a grid where random, normalized gradient vectors are applied at each node. The size of the grid, inputted by the user, dictates the size of the features hence the noise is pseudo-random. This property is ideal for DIC to ensure that the features are large enough for the cameras to see but small enough to improve spatial resolution.
The size of the grid is set at 6 x the projected size of a camera pixel. This ensures non-aliasing (the features should be at least 3 pixels to prevent aliasing) and creates small enough spots.
While Perlin Noise creates smooth textures, a filter is applied to generate black- and white-only features with a 1:1 ratio. This further improves spatial resolution of the measurement by preventing large areas of one color

This simple code can generate a homogenous pseudo-random pattern with optimal feature sizes for DIC applications.
The pattern was applied using paint and a laser-cutted stencil (The MATLAB code provides the files for the laser cutter).
A loop can be added to the MATLAB code to generate multiple results so the user can select the pattern with the minimum number of shapes to simplify the creation of a stencil.
The pattern was successfully used on a 120 x 20 cm structure with reconstruction errors in the order of 0.1 mm.