Initial commit

Flexbrdf/hytools/transform/__init__.py

@@ -0,0 +1,22 @@
+# -*- coding: utf-8 -*-
+"""
+HyTools:  Hyperspectral image processing library
+Copyright (C) 2021 University of Wisconsin
+
+Authors: Adam Chlus, Zhiwei Ye, Philip Townsend.
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, version 3 of the License.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+Transform functions
+"""
+from .resampling import *

Flexbrdf/hytools/transform/resampling.py

@@ -0,0 +1,132 @@
+# -*- coding: utf-8 -*-
+"""
+HyTools:  Hyperspectral image processing library
+Copyright (C) 2021 University of Wisconsin
+
+Authors: Adam Chlus, Zhiwei Ye, Philip Townsend.
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, version 3 of the License.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+Spectral resampling functions.
+
+"""
+import numpy as np
+from scipy.interpolate import interp1d
+
+
+def gaussian(x,mu,fwhm):
+    """
+    Args:
+        x (numpy.ndarray): Values along which to generate gaussian..
+        mu (float): Mean of the gaussian function..
+        fwhm (float): Full width half maximum..
+
+    Returns:
+        numpy.ndarray: Gaussian along input range.
+
+    """
+
+    c = fwhm/(2* np.sqrt(2*np.log(2)))
+    return np.exp(-1*((x-mu)**2/(2*c**2)))
+
+def calc_resample_coeffs(in_wave,in_fwhm,out_wave,out_fwhm, spacing = 1):
+    """Given a set of source and destination wavelengths and FWHMs this
+    function caculates the relative contribution or each input wavelength
+    to the output wavelength. It assumes that both input and output
+    response functions follow a gaussian distribution.
+
+    All inputs shoud be provide in nanometers.
+
+
+    Args:
+        in_wave (list): Input wavelength centers.
+        in_fwhm (list): Input full width half maxes.
+        out_wave (list): Output wavelength centers.
+        out_fwhm (list): Output full width half maxes.
+        spacing (int, optional): Resolution at which to model the
+                    spectral response functions. Defaults to 1.
+
+    Returns:
+        numpy.ndarray: Transform coeffiecients.
+
+    """
+
+    out_matrix = []
+    min_spectrum = min(out_wave.min(),in_wave.min())//100 *100 - 100
+    max_spectrum = 100 + max(out_wave.max(),in_wave.max())//100 *100
+    one_nm = np.arange(min_spectrum,max_spectrum,spacing)
+
+    for wave,fwhm, in zip(out_wave,out_fwhm):
+        a =  gaussian(one_nm,wave,fwhm)
+        out_matrix.append(np.divide(a,np.sum(a)))
+    out_matrix = np.array(out_matrix)
+
+    # For each source wavelength generate the gaussion response
+    in_matrix = []
+    for wave,fwhm in zip(in_wave,in_fwhm):
+        in_matrix.append(gaussian(one_nm ,wave,fwhm))
+    in_matrix = np.array(in_matrix)
+
+    # Calculate the relative contribution of each source response function
+    ratio =  in_matrix/in_matrix.sum(axis=0)
+    ratio[np.isnan(ratio)] = 0
+    ratio2 = np.einsum('ab,cb->acb',ratio,out_matrix)
+
+    # Calculate the relative contribution of each input wavelength
+    # to each destination wavelength
+    coeffs = np.trapz(ratio2)
+
+    return coeffs
+
+
+def apply_resampler(hy_obj,data):
+    ''' Apply SCSS correction to a slice of the data
+
+    Args:
+        hy_obj (TYPE): DESCRIPTION.
+        band (TYPE): DESCRIPTION.
+        index (TYPE): DESCRIPTION.
+
+    Returns:
+        band (TYPE): DESCRIPTION.
+
+    '''
+
+    interp_types = ['linear', 'nearest', 'nearest-up',
+                   'zero', 'slinear', 'quadratic',
+                   'cubic']
+
+    #Convert to float
+    data = data.astype(np.float32)
+
+    if hy_obj.resampler['type'] == 'gaussian':
+
+        # Load resampling coeffs to memory if needed
+        if 'resample_coeffs' not in hy_obj.ancillary.keys():
+            in_wave = hy_obj.wavelengths[~hy_obj.bad_bands]
+            in_fwhm =hy_obj.fwhm[~hy_obj.bad_bands]
+            resample_coeffs = calc_resample_coeffs(in_wave,in_fwhm,
+                                             hy_obj.resampler['out_waves'],
+                                             hy_obj.resampler['out_fwhm'])
+            hy_obj.ancillary['resample_coeffs'] = resample_coeffs
+
+        data = np.dot(data, hy_obj.ancillary['resample_coeffs'] )
+
+
+    elif hy_obj.resampler['type'] in interp_types:
+        interp_func =  interp1d(hy_obj.wavelengths[~hy_obj.bad_bands], data,
+                                kind=hy_obj.resampler['type'],
+                                axis=2, fill_value="extrapolate")
+        data = interp_func(hy_obj.resampler['out_waves'])
+
+    return data