How the algorithm works?

classic Classic list List threaded Threaded
3 messages Options
Batto Batto
Reply | Threaded
Open this post in threaded view
|

How the algorithm works?

Hi everyone!
I try to explain my problem.
I have a 320x320 picture with some particles. If I apply an interrogation window of 64x64 with step 50%, I´ll expect that the result is a 10x10 matrix. Instead, the dimension of matrix result is 9x9.
I have tried with many different images sizes and different interrogations windows, but the result is ever the same (a row and a column less that I expect).
Why are they delete and, in particular, which column and row are delete?
Thanks to those who answer me!
William William
Reply | Threaded
Open this post in threaded view
|

Re: How the algorithm works?

Hi,
Thanks for the feedback. This is normal behaviour, and I try to explain :
When you use 32 px interrogation area and 32px step, then you get what you expect: a 10x10 result matrix.
But when you increase the size of the interrogation area, then the tool will place your first 64x64 area in the corner, and then start to repeat that with 32px steps (overlap). Otherwise, the interrogation areas in the corners would contain 25% particle data only, the rest would be NaN.
The positions of the interrogation areas are determined in piv_fftmulti.m in line 35-58.
This is difficult to explain however...

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.

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

Re: How the algorithm works?

Hi William,
Thank you for your answer. You are totally right, now I have understood.
Thanks a lot!
Cheers!