import os
import numpy as np
from polsartools.utils.proc_utils import process_chunks_parallel
from polsartools.utils.utils import conv2d,time_it
from .dcp_infiles import dcpt2files
[docs]
@time_it
def mf3cd(infolder, window_size=1, outType="tif",
cog_flag=False, cog_overviews = [2, 4, 8, 16],
write_flag=True, max_workers=None,block_size=(512, 512),
progress_callback=None, # for QGIS plugin
):
"""
Computes decomposed powers, and a scattering type parameter from the input dual-co-polarization T2 matrix data, and writes
the output in various formats (GeoTIFF or binary). The computation is performed in parallel for efficiency.
Example:
--------
>>> mf3cd("path_to_T2_folder", window_size=5, outType="tif", cog_flag=True)
This will compute decomposed powers from the dual-co-pol T2 matrix in the specified folder,
generating output in Geotiff format with Cloud Optimized GeoTIFF settings enabled.
Parameters:
-----------
infolder : str
Path to the input folder containing T2 matrix data.
window_size : int, optional
Size of the processing window (default is 1).
outType : str, optional
Output format of the files; can be "tif" (GeoTIFF) or "bin" (binary) (default is "tif").
cog_flag : bool, optional
If True, outputs Cloud Optimized GeoTIFF (COG) (default is False).
cog_overviews : list of int, optional
List of overview levels to be used for COGs (default is [2, 4, 8, 16]).
write_flag : bool, optional
Whether to write the computed output files (default is True).
max_workers : int, optional
Number of parallel workers for processing (default is None, which uses one less than the number of available CPU cores).
block_size : tuple of int, optional
Size of each processing block (default is (512, 512)), defining the spatial chunk dimensions used in parallel computation.
Returns:
--------
None
The function writes the computed decomposed powers to the specified output format.
Output Files:
-------------
- "Ps_mf3cd.tif" or "Ps_mf3cd.bin"
- "Pd_mf3cd.tif" or "Pd_mf3cd.bin"
- "Pv_mf3cd.tif" or "Pv_mf3cd.bin"
- "Theta_DP_mf3cd.tif" or "Theta_DP_mf3cd.bin"
"""
input_filepaths = dcpt2files(infolder)
output_filepaths = []
if outType == "bin":
output_filepaths.append(os.path.join(infolder, "Ps_mf3cd.bin"))
output_filepaths.append(os.path.join(infolder, "Pd_mf3cd.bin"))
output_filepaths.append(os.path.join(infolder, "Pv_mf3cd.bin"))
output_filepaths.append(os.path.join(infolder, "Theta_DP_mf3cd.bin"))
else:
output_filepaths.append(os.path.join(infolder, "Ps_mf3cd.tif"))
output_filepaths.append(os.path.join(infolder, "Pd_mf3cd.tif"))
output_filepaths.append(os.path.join(infolder, "Pv_mf3cd.tif"))
output_filepaths.append(os.path.join(infolder, "Theta_DP_mf3cd.tif"))
process_chunks_parallel(input_filepaths, list(output_filepaths),
window_size=window_size, write_flag=write_flag,
processing_func=process_chunk_mf3cd,
block_size=block_size, max_workers=max_workers,
num_outputs=len(output_filepaths),
cog_flag=cog_flag,
cog_overviews=cog_overviews,
progress_callback=progress_callback
)
def process_chunk_mf3cd(chunks, window_size,*args):
kernel = np.ones((window_size,window_size),np.float32)/(window_size*window_size)
t11_T1 = np.array(chunks[0])
t12_T1 = np.array(chunks[1])+1j*np.array(chunks[2])
t21_T1 = np.conj(t12_T1)
t22_T1 = np.array(chunks[3])
if window_size>1:
t11s = conv2d(np.real(t11_T1),kernel)+1j*conv2d(np.imag(t11_T1),kernel)
t12s = conv2d(np.real(t12_T1),kernel)+1j*conv2d(np.imag(t12_T1),kernel)
t21s = np.conj(t12_T1)
t22s = conv2d(np.real(t22_T1),kernel)+1j*conv2d(np.imag(t22_T1),kernel)
else:
t11s = t11_T1
t12s = t12_T1
t21s = t21_T1
t22s = t22_T1
det_T2 = t11s*t22s-t12s*t21s
trace_T2 = t11s + t22s
m1 = np.real(np.sqrt(1-(4*(det_T2/(trace_T2**2)))))
h = (t11s - t22s)
g = t22s
span = t11s + t22s
val = (m1*span*h)/(t11s*g+m1**2*span**2);
thet = np.real(np.arctan(val))
# thet = np.rad2deg(thet)
theta_DP = np.rad2deg(thet)
Ps_DP = (((m1*(span)*(1+np.sin(2*thet))/2)))
Pd_DP = (((m1*(span)*(1-np.sin(2*thet))/2)))
Pv_DP = (span*(1-m1))
return np.real(Ps_DP),np.real(Pd_DP),np.real(Pv_DP),np.real(theta_DP)
