Complex numbers in PIV processing

classic Classic list List threaded Threaded
6 messages Options
Kasia M. Kasia M.
Reply | Threaded
Open this post in threaded view
|

Complex numbers in PIV processing

Hello!

First, I would like to thank you for making PIVlab such a user-friendly and efficient PIV processing tool.

I use PIVlab for determining the velocities induced by a relatively slow convective flow, and am likewise running into problems with the appearance of complex numbers for some of my experiments. I pre-process my images in Fiji, and define a ROI as well as create a mask for additional regions in PIVlab v1.35. For PIV settings, I use:

FFT window deformation(128,64, and 32 px^2 with 50% step overlap)
A linear interpolator
A 2D Gauss sub-pixel estimator

I process 100 image pairs, and some of the resulting velocity component values are complex (however, not for all pairs). Please let me know if you would like me to provide you with images for easier diagnosis, and if you have more questions. Thank you so much!

Best,
Kasia
William William
Reply | Threaded
Open this post in threaded view
|

Re: Complex numbers in PIV processing

Hi Kasia,
I think I did not fix the complex number bug in the 2D gauss subpix estimator yet. So a workaround would be to use the standard 2x3 Point estimator. That one has a very similar performance when using the FFT window deformation algorithm.

Please consider donating for the PIVlab project:
http://paypal.me/pools/c/81Y8wBVi4y

The developer needs to update to a recent Matlab release in order to keep the project running.

Kasia M. Kasia M.
Reply | Threaded
Open this post in threaded view
|

Re: Complex numbers in PIV processing

Hi William,

Thank you for your reply. I tried processing with the 2x3 point estimator, and some of my frames still contain complex numbers. It is not a big problem, because I can find the mean velocity fields without using the few complex frames, but I did want to let you know for your diagnostic purposes. Thanks!

-Kasia
Stephan C Stephan C
Reply | Threaded
Open this post in threaded view
|

Re: Complex numbers in PIV processing

Same issue encountered here while performing DCC, so has nothing to do with the FFT processing. Subpixel est was the 2D gauss.
William William
Reply | Threaded
Open this post in threaded view
|

Re: Complex numbers in PIV processing

yes, it is a bug in the subpixel estimator. This bug was fixed for the 2x3-point estimator in FFT algorithm as far as I can say, but not yet for 2D gauss or DCC algorithms.

Please consider donating for the PIVlab project:
http://paypal.me/pools/c/81Y8wBVi4y

The developer needs to update to a recent Matlab release in order to keep the project running.

William William
Reply | Threaded
Open this post in threaded view
|

Re: Complex numbers in PIV processing

This bug should now be fixed in PIVlab v1.4

Please consider donating for the PIVlab project:
http://paypal.me/pools/c/81Y8wBVi4y

The developer needs to update to a recent Matlab release in order to keep the project running.