µDIC: Digital Image Correlation Toolkit

µDIC is a 'batteries included' Python toolkit for Digital Image Correlation (DIC), designed for both experimental data analysis and virtual experiments. The toolkit is organized into several specialized packages:

• IO-Tools: For handling input and output operations.
• Virtual Lab: Tools for simulating experimental conditions, including:
  • Speckle generation
  • Image deformation
  • Noise injection
  • Image downsampling
• Mesh generation: Includes a lightweight GUI for structured meshing and support for B-spline elements (with arbitrary polynomial order and manipulatable knot vectors).
• Image correlation routines: Implements a non-linear least squares solver.
• Post processor: Calculates common strain measures and provides lightweight visualization tools.
• Test suites: Automated function tests to ensure reliability.

µDIC is a "batteries included" Python toolkit for Digital Image Correlation (DIC), designed for both experimental data analysis and virtual experiments. It is implemented in pure Python (relying on Scipy and Numpy) to ensure transparency and ease of extension.

Key components include:

• Image reader tools: For loading experimental data.
• Virtual lab: Tools for speckle image generation, image deformation, noise injection, and down-sampling.
• B-spline finite elements: Supports arbitrary polynomial orders and manipulatable knot vectors.
• Meshing tools: Includes a lightweight GUI for structured meshing.
• Image correlation routines: Uses a non-linear least squares solver.
• Post processor: Calculates common strain measures and provides lightweight visualization (including quiver plots for displacements).

The Virtual lab is a collection of tools designed to simulate virtual tensile tests on speckle patterns. It allows you to combine various effects to create realistic synthetic datasets for Digital Image Correlation (DIC) testing. You can chain together the following capabilities:

• Speckle Generation: Create patterns using Perlin noise or drop spray methods, with additional tools for blurring.
• Downsampling: Simulate sensor resolution limits using bi-cubic spline interpolation. Supports configurable downsampling factors, fill-factors, and random pixel offsets.
• Noise Injection: Add Gaussian noise to simulate sensor artifacts.
• Image Deformation: Apply arbitrary deformation gradients or complex deformation gradient fields.
• Virtual Test Orchestration: Use the VirtualTest class to combine all these features into a single workflow.

Digital Image Correlation (DIC) is a method used to measure the deformation of an object by analyzing a sequence of images captured during the deformation process.

Instead of tracking individual points manually, DIC relies on the conservation of optical flow. This fundamental assumption states that the grayscale values of pixels in an initial image ($g$ at position $\boldsymbol{x}_0$) are preserved and simply convected to new positions ($\boldsymbol{x}$) in a subsequent image ($f$).

Mathematically, the goal is to find $\boldsymbol{x}$ such that:

$$g(\boldsymbol{x}_0) = f(\boldsymbol{x})$$

Because the new positions $\boldsymbol{x}$ are often not aligned with the center of a pixel, the routine must use interpolation to determine the grayscale values of $f$ at those non-integer coordinates.

The ImageDeformer class and its associated functions in muDIC.vlab.image_deformer allow you to apply controlled deformations to images to simulate physical strain or movement.

Key ways to initialize a deformer include:

• imageDeformer_from_defGrad: Create a deformer based on a specific deformation gradient.
• imageDeformer_from_uFunc: Create a deformer based on a user-defined function (uFunc) that describes the displacement field.

Related utility functions:

• deformation_gradient: Computes the deformation gradient from displacement data.
• num_diff: Provides numerical differentiation capabilities required for deformation calculations.

Sneeze is an algorithm used to generate synthetic speckle patterns for Digital Image Correlation (DIC). It simulates a spray-painting process by iteratively adding random ink drops to a canvas.

The algorithm follows these steps:

1. Initializes an empty canvas.
2. Generates a random ink drop radius and location.
3. Appends the drop to the canvas.
4. Repeats the process for the specified number of drops.
5. Applies a Gaussian blur to the canvas to round the droplets and simulate realistic speckle morphology.

If you already have a list of image objects loaded in memory, you can convert them into a µDIC image stack using dic.image_stack_from_list(list_of_images).

import muDIC as dic

# Assuming list_of_images is already defined
image_stack = dic.image_stack_from_list(list_of_images)

Use dic.image_stack_from_folder(path, filetype=".png") to load all images of a specific type from a directory into a µDIC image stack. The path should be the directory containing the images, and filetype specifies the extension to look for.

import muDIC as dic

path = r"/the/path/to/the/images/"
image_stack = dic.image_stack_from_folder(path, filetype=".png")

Once you have configured your DIC_settings, you execute the analysis by creating a DIC_job object and calling its .run() method. The results of the entire analysis are returned directly by the .run() method.

import muDIC as dic

# Assuming settings is already configured
dic_job = dic.DIC_job(settings)
results = dic_job.run()