How to find good PIV parameters?

Fluidimage is designed to process several images in parallel with asynchronous “topologies”. However, the asynchronous PIV topology is not very convenient to look for the best parameters for a PIV computation. It is better to use the PIV “work” directly as in these examples:

"""To be run in IPython to find a good set of parameters"""

from fluidimage.piv import Work

params = Work.create_default_params()

params.series.path = "../../image_samples/Oseen/Images"

params.mask.strcrop = "30:250, 40:"

params.piv0.shape_crop_im0 = 32
params.piv0.displacement_max = 7

params.fix.correl_min = 0.2
params.fix.threshold_diff_neighbour = 8

params.multipass.number = 2
params.multipass.use_tps = "last"

work = Work(params=params)

piv = work.process_1_serie()

# piv.display(show_interp=True, scale=0.3, show_error=True)
piv.display(show_interp=False, scale=1, show_error=True)

# piv.save()

The parameters in params.series are used to define a fluiddyn.util.serieofarrays.SeriesOfArrays and to select one serie (which represents here a couple of images). It is also what is done internally in the PIV topology. Have a look at our tutorial to discover how to use this powerful tool!

This other example includes a simple image-to-image preprocessing.

from fluidimage.image2image import apply_im2im_filter
from fluidimage.piv import Work

params = Work.create_default_params()

params.multipass.number = 2
params.multipass.use_tps = True

params.piv0.shape_crop_im0 = 32
params.piv0.displacement_max = 5
params.fix.correl_min = 0.2
params.fix.threshold_diff_neighbour = 8

params.mask.strcrop = "30:250, 100:"

path = "../../image_samples/Oseen/Images"
# path = '../../image_samples/Karman/Images'
params.series.path = path
params.series.str_subset = "i+1:i+3"

work = Work(params=params)

serie = work.get_serie()

# "image to image" filter
serie = apply_im2im_filter(serie, im2im="my_example_im2im.im2im")

piv = work.calcul(serie)

# piv.display(show_interp=True, scale=0.3, show_error=True)
piv.display(show_interp=False, scale=1, show_error=True)

# piv.save()