examples.rst

F3D offers the following as usable filters:

• Median Filter
• Bilateral Filter
• Mask Filter
• MM Filter: Opening
• MM Filter: Closing
• MM Filter: Dilation
• MM Filter: Erosion

A FFT filter will be implemented in the future.

To use the filters, you may either use built-in functions for single filters, or create a list of filters. For example:

import pyF3D as f
import tifffile as t

image = t.imread('.../example.tif')
median = f.run_MedianFilter(image)

This code will run a median filter on the image example.tif. To use multiple filters, create a list of filters while filling out optional parameters for each filter:

pipeline = [f.MedianFilter(), f.MMFilterEro(mask='Diagonal3x3x3')]
new_image = f.run_f3d(image, pipeline)

This code will run a median filter, then an erosion filter, on the image. You may also specify the platforms on which to run the calculations. F3D provides a helper function pyF3D.list_all_cl_platforms, which will specify all platforms you can access for processing. By default, F3D will use the first device on this list. There are several ways to specify the platforms to use:

1. Specify a single OpenCL platform in an F3D function's arguments:

   platform = f.list_all_cl_platforms()[0]
   new_image = f.run_f3d(image, pipline, platform=platform)

2. Specify a list of platforms, for processing in parallel:

   platforms = f.list_all_cl_platforms()
   new_image = f.run_f3d(image, pipline, platform=platforms)

3. Specify a dictionary of platform/int key-value pairs. The int value specifies the maximum number of slices to be loaded onto the corresponding platform.

   platforms = {}
   for p in f.list_all_cl_platforms():
       platforms[p] = 200
   new_image = f.run_f3d(image, pipline, platform=platforms)

   This code ensures that no more than 200 slices of the image will be loaded onto each platform at any time.