Speckle quality and ARAMIS

[[Si...]]

Since DIC is starting to mature it is becoming increasingly important to report more information about the test setup and parameters used in for example journals. Speckle size and contrast is one of these areas and it is well known that a "speckle" need to be 3-5 pixels (I think ARAMIS wants 5?). This is straight forward with a speckle pattern consisting of perfect black dots on a white background, but when using DIC and metal tensile specimens a randomised speckle pattern of black and white spray paint is often used which leaves a considerable amount of gray in the image spectrum. ARAMIS has not complained about this and facets are not lost even for rather small facets. The question I have is how ARAMIS determine the "size" of a speckle, i.e. the prominence level that need to be reached with regard to the surrounding pixels for it to be acceptable. Is any filtering done (for example FFT bandpass filtering) before calculations are made to increase contrast?

I understand that this might be borderline intellectual property secrets but I thought I would ask anyway since it is rather important in academia. There are several techniques to obtain speckle sizes, but if it is not the same parameters as the software uses it feels rather useless in trying to quantifying it.

Any clarification would be greatly appreciated!

 

Thank you in advance!

[[Iv...]]

Dear Simon,

the "speckle pattern" is also known as stochastic pattern. The target of the whole pattern is to have stochastic distribution of different white and black pixels. Those pixel receive a value of it's brightness or darkness, so called grayvalue. 

To create a surface or 3D-coordinates you use facets. The facets are squares with a size you can define in the creation dialog. To find the same facet in an other image or after the movement or deformation the software checks the grayvalue of each pixel within the facet and search for the nearly same distribution in the following image. 

So you need a completely different distribution of grayvalues. A certain distribution with 3 pixels white and 3 pixel dark and so on doesn't create a unique facet and so it's impossible to find the same facet all over again because an other facet can be to similar. Of course you have to consider that you can't have such big speckles that your whole facet is all black or white. 

I recommend to download the ARAMIS App from the Appstore. There is a pattern generator implemented which shows you a example of you distribution for different measuring volumes. 

But after all, if you create a pattern stochastic and so as unique as possible you receive better values. 

Greetings, Ivan