Polarimetric Refined Lee speckle filter¶

This tutorial shows how to:

• Open the ALOS dataset and convert the S matrix to T3 coherency matrix
• Pre-process the data by applying a 4-look decimation in azimuth
• Apply the Refined Lee speckle filter to the 3x3 coherency matrix
• Write the results to a file
• Visualize the results as an RGB image

In [1]:

%load_ext autoreload
%autoreload 2

import os

import xarray as xr

from pathlib import Path
from polsarpro.io import open_netcdf_beam, polmat_to_netcdf
from polsarpro.speckle_filters import refined_lee
from polsarpro.util import S_to_T3, multilook, pauli_rgb

# optional import for progress bar
from dask.diagnostics import ProgressBar


# change to your data paths
# original dataset
input_alos_data = Path("/data/psp/test_files/SAN_FRANCISCO_ALOS1_slc.nc")

# output files from C
output_test_dir = Path("/data/psp/res/")

%load_ext autoreload %autoreload 2 import os import xarray as xr from pathlib import Path from polsarpro.io import open_netcdf_beam, polmat_to_netcdf from polsarpro.speckle_filters import refined_lee from polsarpro.util import S_to_T3, multilook, pauli_rgb # optional import for progress bar from dask.diagnostics import ProgressBar # change to your data paths # original dataset input_alos_data = Path("/data/psp/test_files/SAN_FRANCISCO_ALOS1_slc.nc") # output files from C output_test_dir = Path("/data/psp/res/")

Load ALOS data¶

In [2]:

S = open_netcdf_beam(input_alos_data)

S = open_netcdf_beam(input_alos_data)

Apply the filter to T3 matrices¶

In [ ]:

if os.path.exists(output_test_dir / "T3_refined_lee.nc"):
    os.remove(output_test_dir / "T3_refined_lee.nc")

with ProgressBar():
    # convert S to T3
    T3 = S_to_T3(S)
    # apply 4-look multilook along lines
    T3 = multilook(T3, dim_az=4, dim_rg=1)
    # apply the filter
    T3_flt = refined_lee(T3, window_size=7, num_looks=4)
    # use custom function that writes complex matrices and preserves chunks
    polmat_to_netcdf(T3_flt, output_test_dir / f"T3_refined_lee.nc")
    

if os.path.exists(output_test_dir / "T3_refined_lee.nc"): os.remove(output_test_dir / "T3_refined_lee.nc") with ProgressBar(): # convert S to T3 T3 = S_to_T3(S) # apply 4-look multilook along lines T3 = multilook(T3, dim_az=4, dim_rg=1) # apply the filter T3_flt = refined_lee(T3, window_size=7, num_looks=4) # use custom function that writes complex matrices and preserves chunks polmat_to_netcdf(T3_flt, output_test_dir / f"T3_refined_lee.nc")

[########################################] | 100% Completed | 27.47 s

Open the file and visualize the results as an RBG image¶

In [ ]:

# open the file
T3_flt = open_netcdf_beam(output_test_dir / f"T3_refined_lee.nc")

pauli_rgb(T3_flt).plot.imshow(figsize=(5, 7))

# open the file T3_flt = open_netcdf_beam(output_test_dir / f"T3_refined_lee.nc") pauli_rgb(T3_flt).plot.imshow(figsize=(5, 7))

Out[ ]:

<matplotlib.image.AxesImage at 0x7a68100504a0>