Why Do Identical Images Yield Different Strain Results?

When analyzing data with a DIC system, users may find:

With the same set of raw images, changing only the calculation parameters changed the final maximum strain, the smoothness of contour maps and the local strain distribution.

This does not necessarily mean the software calculation is unreliable.

DICsystemsStrain is not read directly from a single pixel; displacement is first computed from image features within a certain range, and then strain is computed from the spatial variation of the surrounding displacement.

Therefore, the subset, calculation step size, strain window and filtering method all affect the spatial scale at which the system observes deformation.

What Is a DIC Subset?

DIC software usually divides the measurement area into many small image areas, each called a subset.

The subset contains a certain number of speckle grayscale features. The software determines the displacement of that area by comparing the feature changes of the same subset in the reference and deformed images.

What Happens If the Subset Is Too Small?

If the subset is too small, there may not be enough speckle features inside, causing:

▪ Unstable image matching;

▪ Increased calculation noise;

▪ Correlation failure in local areas;

▪ Scattered abnormal points appear in the contour map.

▪ What happens if the subset is too large?

If the subset is too large, although matching stability is usually improved, it may also:

▪ Averaged local deformation;

▪ Reduced spatial resolution;

▪ Reducing the peaks near cracks or notches;

▪ Blur the boundaries between different deformation areas.

Therefore, the subset is not the smaller the better, nor the larger and more stable the better.

What Is the Calculation Step Size?

The calculation step size represents the distance between adjacent calculation points.

A smaller step size increases the density of calculation points, making contour maps appear more continuous; a larger step size reduces the computation load and improves analysis efficiency.

But a small calculation step does not necessarily mean higher real spatial resolution.

If the subset itself covers a large area, even if adjacent calculation points are densely arranged, different results still come from largely overlapping image areas.

Therefore, a large number of calculation points does not mean each point corresponds to completely independent physical information.

What Is the Strain Window?

DICFirst the displacement field is obtained, then strain is computed from the displacement changes of neighboring areas.

The strain window determines how large a range of displacement data is referenced when computing a strain result.

A smaller strain window may appear as:

▪ More sensitive to local changes;

▪ Higher strain peaks;

▪ More obvious gradients;

▪ Random noise is also more easily amplified.

A larger strain window may appear as:

▪ Smoother contour maps;

▪ Reduced random fluctuation;

▪ Reduced local peaks;

▪ Small-area strain concentration is averaged.

This is why the same set of displacement data may yield different strain contour maps under different strain windows.

Is a Smoother Strain Contour Map Always Better?

Not necessarily.

Overly rough contour maps may contain much noise, but over-smoothing may also mask real local deformation.

In a uniform tension area, reasonable strain results should usually have good continuity. If the contour map shows randomly distributed high/low value points, check the speckle, lighting and parameter settings.

But near holes, notches, weld seams and cracks, strain can change rapidly in the first place. Blindly pursuing smoothness may reduce local strain analysis capability.

Therefore, judging whether parameters are reasonable cannot rely only on whether the contour map looks good; one must judge whether the results match the specimen structure, loading mode and actual deformation laws.

How to Choose Appropriate DIC Parameters?

▪ Select subsets according to speckle

The subset needs to contain a sufficient number of randomly distributed speckle features.

Speckle that is too large, too small, insufficient in contrast or too repetitive in distribution can all affect the relevant calculations.

Therefore, calculation parameters cannot be determined separately from the original images.

▪ Determine the spatial scale according to the target size

If what is studied is the overall uniform deformation of the specimen, calculation stability can be appropriately increased.

If you study hole edges, crack tips, microscopic specimens or local weld deformation, more attention should be paid to spatial resolution to avoid over-averaging of local results.

▪ Evaluate noise using static images

When the specimen has no real deformation, continuous static images can be captured for calculation to observe displacement and strain fluctuation.

If large random strain already appears in the static state, it indicates that the current imaging conditions or calculation parameters may not be suitable for direct local deformation judgment.

▪ Compare the stability after parameter changes

Parameters can be adjusted within a reasonable range to observe whether the positions of the main deformation areas are stable.

Real strain concentration usually keeps a similar position and development trend under different parameters, while random noise tends to move noticeably.

Why Not Just Compare the Maximum Strain Values?

The maximum strain usually comes from a very small local area and is susceptible to the following factors:

▪ Image noise;

▪ Calculation boundaries;

▪ Local reflections;

▪ Missing speckle;

▪ Crack opening;

▪ Strain window size.

In material performance evaluation, besides maximum values, one should also observe:

▪ The location of high-strain areas;

▪ The extent of high-strain areas;

▪ Regional average strain;

▪ Strain changes along specified paths;

▪ The development trend of strain with load;

▪ Whether results are stable under different parameters.

This avoids drawing wrong conclusions from a single abnormal point.

Which Parameters Should Be Retained in a DIC Test Report?

To ensure reproducibility, a DIC test report should not only provide the final contour maps but also state:

▪ Measurement field of view；

▪ Image resolution;

▪ Speckle preparation method;

▪ Subset size;

▪ Computation step size;

▪ Strain window;

▪ Filtering method;

▪ Camera sampling frequency;

▪ 2D or 3D measurement modes;

▪ System calibration and test setup conditions.

Without this information, even with the same set of images, it may be difficult to reproduce the same results.

How Does Haytham Handle DIC Parameter Configuration Issues?

There is no fixed template of DIC calculation parameters applicable to all tests.

The same subset and strain window may be relatively stable for uniform tension, but for microscopic specimens, crack tips, large deformation or complex structures, they may not balance spatial resolution and calculation stability.

Haytham TechnologyIn DIC project configuration, the following will be combined:

▪ Specimen dimensions;

▪ Measurement field of view；

▪ Speckle pixel size;

▪ Target strain range;

▪ Local feature size;

▪ Deformation speed;

▪ Result analysis objectives.

▪ Match image acquisition, optical configuration and calculation parameters.

Haytham's DIC visual strain measurement software analyzes full-field displacement, strain, regional averages and data along specified paths, with measurement areas and calculation methods configurable for different test objectives.

For enterprise quality verification, research papers and multi-batch material comparisons, establishing unified imaging condition and calculation parameter templates is more meaningful than merely pursuing a smoother contour map or a higher single peak.

Frequently Asked Questions

▪ Does a smaller subset mean higher DIC measurement accuracy?

Not necessarily. A too-small subset may lack sufficient image features, increasing calculation noise or causing correlation failure. The subset needs to match the speckle size and image quality.

▪ Does a smaller calculation step mean higher spatial resolution?

Not entirely. A smaller step size increases calculation point density, but real spatial resolution is still limited by the subset, field of view and pixel count.

▪ Why does the peak decrease after enlarging the strain window?

A larger strain window computes displacement gradients over a wider range, so local peaks may be averaged and contour maps usually become smoother.

▪ Which set of parameters is the correct one?

There are no universal parameters for all tests. Reasonable parameters should ensure stable image correlation and controllable noise while retaining the local details required by the research objective.