def get_padded_feature_stack(window_size, correction_switch, degree, pad=-1):
eigen_full=extract_polarimetric.eigen_raster_full(window_size, correction_switch, degree) #9 is the window size
cov_arr=extract_polarimetric.extract_covariance_arr(window_size, correction_switch, degree)
#padding=int(np.floor(window_size/2))
padding=window_size//2 if pad!=0 else 0
Ihh=np.pad(cov_arr[...,0,0], padding, 'constant')
Ihv=np.pad(cov_arr[...,1,1], padding, 'constant')
Ivv=np.pad(cov_arr[...,2,2], padding, 'constant')
arr_lamb1=np.pad(eigen_full[:,:,2], padding, 'constant')
arr_lamb2=np.pad(eigen_full[:,:,1], padding, 'constant')
arr_lamb3=np.pad(eigen_full[:,:,0], padding, 'constant')
co_pol_dif=np.pad(extract_polarimetric.co_pol_diff(cov_arr), padding, 'constant')
arr_det_cov=np.pad(np.absolute(extract_polarimetric.determinant_cov_conj(cov_arr)), padding, 'constant')
R_co = Rco_X = np.pad(np.real(cov_arr[...,0,2]), padding, 'constant')
Ico_X = np.pad(np.absolute(np.imag(cov_arr[...,0,2])), padding, 'constant')
#return np.dstack((np.real(Ihh),np.real(Ihv),np.real(Ivv),np.real(arr_lamb1), np.real(arr_lamb2), np.real(arr_lamb3), np.absolute(co_pol_dif),R_co, Ico_X, np.absolute(arr_det_cov)))
return np.dstack((np.real(Ihh),np.real(Ihv),np.real(Ivv),np.real(arr_lamb1), np.real(arr_lamb2), np.real(arr_lamb3), np.absolute(co_pol_dif), np.absolute(arr_det_cov)))#,R_co, Ico_X
def krogager_array(window_size, correction, degree):
cov_arr=extract_polarimetric.extract_covariance_arr(window_size, correction, degree)
clip_extremes=True
ShhShh=cov_arr[:,:,0,0]
SvvSvv=cov_arr[:,:,2,2]
ShvShv=cov_arr[:,:,1,1]
ShhSvv_=cov_arr[:,:,0,2]
Shh_Svv=cov_arr[:,:,2,0]
co_pol_sum_mag=ShhShh+SvvSvv+ShhSvv_+Shh_Svv
co_pol_diff_mag=ShhShh+SvvSvv-ShhSvv_-Shh_Svv
ShvShh_=cov_arr[:,:,1,0]/np.sqrt(2)
ShvSvv_=cov_arr[:,:,1,2]/np.sqrt(2)
Shv_Shh=cov_arr[:,:,0,1]/np.sqrt(2)
Shv_Svv=cov_arr[:,:,2,1]/np.sqrt(2)
iota=1j
Srr_mag_complex=ShvShv+ 0.25*co_pol_diff_mag +iota*(ShvShh_-ShvSvv_-Shv_Shh-Shv_Svv)#jShv+0.5*(Shh-Svv))
Srr_mod=np.absolute(Srr_mag_complex)
Sll_mag_complex=ShvShv+ 0.25*co_pol_diff_mag -iota*(ShvShh_-ShvSvv_-Shv_Shh-Shv_Svv)#jShv-0.5*(Shh-Svv))
Sll_mod=np.absolute(Sll_mag_complex)
Srl_mag=ShhShh+SvvSvv+iota*(ShhSvv_+Shh_Svv)
Srl_mod=np.absolute(Srl_mag)
ks=Srl_mod
#if(Srr_mod>Sll_mod):
kd=np.minimum(Sll_mod, Srr_mod)
kh=np.maximum(Sll_mod, Srr_mod)-kd
#else:
#kd=Srr_mod
#kh=Sll_mod-Srr_mod
return np.dstack((kd,kh,kh)) #mod-> k**2
def plot_feature_space_test(window_size):
eigen_arr = extract_polarimetric.eigen_raster_full(window_size)
cov_arr = extract_polarimetric.extract_covariance_arr(window_size, False)
cppr = np.absolute(extract_polarimetric.co_pol_power_ratio_1(cov_arr))
co_pol_diff = np.absolute(extract_polarimetric.co_pol_diff(cov_arr))
lambda_3 = np.absolute(eigen_arr[:, :, 0])
plot_feature_space(co_pol_diff, lambda_3)
def cloude_pottier(window_size):
fig, ax = plt.subplots(nrows=3, ncols=3)
arr = extract_polarimetric.eigen_raster_full(window_size)
cov_arr = extract_polarimetric.extract_covariance_arr(window_size)
plt.subplot(3, 3, 1)
plt.imshow(arr[:, :, 2], cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('lambda_1')
plt.colorbar()
plt.subplot(3, 3, 2)
plt.imshow(arr[:, :, 1], cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('lambda_2')
plt.colorbar()
plt.subplot(3, 3, 3)
plt.imshow(arr[:, :, 0], cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('lambda_3')
plt.colorbar()
plt.subplot(3, 3, 4)
plt.imshow(extract_polarimetric.entropy(arr), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Entropy')
plt.colorbar()
plt.subplot(3, 3, 5)
plt.imshow(extract_polarimetric.anisotropy(arr), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Anisotropy')
plt.colorbar()
plt.subplot(3, 3, 6)
plt.imshow(extract_polarimetric.pol_fraction(arr), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Pol_fraction')
plt.colorbar()
plt.subplot(3, 3, 7)
plt.imshow(extract_polarimetric.pedestal_height(arr), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('pedestal_height')
plt.colorbar()
plt.show()
def Pauli_RGB_array(window_size, correction):
cov_arr = extract_polarimetric.extract_covariance_arr(
window_size, correction)
Shh = np.real(cov_arr[:, :, 0, 0])
Svv = np.real(cov_arr[:, :, 2, 2])
Shv = np.real(cov_arr[:, :, 1, 1])
b = hist_stretch_all(Shh + Svv, 0)
r = hist_stretch_all(np.absolute(Shh - Svv), 0)
g = hist_stretch_all(Shv, 0)
img_arr = np.dstack((r, g, b))
return img_arr
def Freeman_Durdun_Decomposition_1(window_size, correction, degree):
#, leaping_win=False, stride_row=1, stride_col=1
cov_arr=extract_polarimetric.extract_covariance_arr(window_size, correction, degree)
c11=cov_arr[:,:,0,0]
c22=cov_arr[:,:,1,1]
c33=cov_arr[:,:,2,2]
c13=cov_arr[:,:,0,2]
c13_re=np.real(c13)
c13_im=np.imag(c13)
shp=c11.shape
rows=shp[0]
cols=shp[1]
alpha=beta=fv=fs=fd=np.empty((rows, cols), dtype=np.complex64)
#=c11,c11,c11
#Freeman algo
#fv=1.5*c22
#c11=c11-fv
#c33=c33-fv
#c13_re=c13_re-(fv/3)
stride_row, stride_col=1,1
for i in range(0, rows, stride_row):
for j in range(0, cols, stride_col):
fv[i,j]=1.5*c22[i,j]
c11[i,j]=c11[i,j]-fv[i,j]
c33[i,j]=c33[i,j]-fv[i,j]
c13_re[i,j]=c13_re[i,j]-(fv[i,j]/3)
#single bounce scattering dominates
if(c13_re[i,j]>=0):
#print(i,j)
alpha[i,j]=-1
fd[i,j]=(c11[i,j]*c33[i,j] - c13_re[i,j]**2 - c13_im[i,j]**2)/(c11[i,j]+c33[i,j]-2*c13_re[i,j])
fs[i,j]=c33[i,j]-fd[i,j]
beta[i,j]=np.sqrt((fd[i,j]+c13_re[i,j]) * (fd[i,j]-c13_re[i,j]) + c13_im[i,j]**2)/fs[i,j]
#single bounce scattering dominates
if(c13_re[i,j]<0):
#print(i,j)
beta[i,j]=1
fs[i,j]=(c11[i,j]*c33[i,j] - c13_re[i,j]**2 - c13_im[i,j]**2)/(c11[i,j]+c33[i,j]-2*c13_re[i,j])
fd[i,j]=c33[i,j]-fs[i,j]
alpha[i,j]=np.sqrt((fs[i,j]+c13_re[i,j]) * (fs[i,j]-c13_re[i,j]) + c13_im[i,j]**2)/fd[i,j]
print(i)
Ps=fs*(1+np.absolute(beta)**2)
Pd=fd*(1+np.absolute(alpha)**2)
Pv=8*fv/3
return np.dstack((Pd,Pv,Ps))
def make_cov_arr_df(window_size, correction_switch, degree):
cov_arr = extract_polarimetric.extract_covariance_arr(
window_size, correction_switch, degree)
shp_cov = cov_arr.shape
linear_cov_arr = np.reshape(
cov_arr, (np.prod(shp_cov[:2]), shp_cov[2] * shp_cov[3]))
#print(linear_cov_arr[1])
cov_df = pd.DataFrame(linear_cov_arr,
columns=[
'Ihh', 'ShhShv', 'ShhSvv', 'ShvShh', 'Ihv',
'ShvSvv', 'SvvShh', 'SvvShv', 'Ivv'
])
#cov_df=cov_df.drop(['ShvShh','SvvShh','SvvShv'], axis=1)
return cov_df
def polarimetric_features(window_size):
fig, ax = plt.subplots(nrows=2, ncols=3)
#arr=extract_polarimetric.eigen_raster_full(window_size)
cov_arr = extract_polarimetric.extract_covariance_arr(window_size, False)
plt.subplot(2, 3, 1)
plt.imshow(np.absolute(extract_polarimetric.co_pol_power_ratio_1(cov_arr)),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Ivv/Ihh (Co-pol power ratio)')
plt.colorbar()
plt.subplot(2, 3, 2)
plt.imshow(np.absolute(extract_polarimetric.determinant_cov(cov_arr)),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Det(Cov)')
plt.colorbar()
plt.subplot(2, 3, 3)
plt.imshow(np.absolute(extract_polarimetric.co_pol_diff(cov_arr)),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Co-pol diff (Ihh-Ivv)')
plt.colorbar()
plt.subplot(2, 3, 4)
plt.imshow(np.real(extract_polarimetric.co_pol_cross_product(cov_arr)),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Real(ShhSvv) co-pol cross_product')
plt.colorbar()
plt.subplot(2, 3, 5)
plt.imshow(np.imag(extract_polarimetric.co_pol_cross_product(cov_arr)),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Imag(ShhSvv) Co-pol cross_product')
plt.colorbar()
plt.show()
def read_Pol_features(window_size, correction_switch, degree, rescale_switch = False, log_Transform_switch = False):
#os.chdir('../')
eigen_full = extract_polarimetric.eigen_raster_full(window_size, correction_switch, degree) #9 is the window size
cov_arr = extract_polarimetric.extract_covariance_arr(window_size, correction_switch, degree)
arr_lamb1=np.absolute(eigen_full[:,:,2])
arr_lamb2=np.absolute(eigen_full[:,:,1])
arr_lamb3=np.absolute(eigen_full[:,:,0])
co_pol_dif=np.absolute(extract_polarimetric.co_pol_diff(cov_arr))
arr_det_cov=np.absolute(extract_polarimetric.determinant_cov_conj(cov_arr))
Rco_X = np.real(cov_arr[...,0,2])
Ico_X = abs(np.imag(cov_arr[...,0,2]))
I_hh = np.absolute(cov_arr[...,0,0])
I_hv = np.absolute(cov_arr[...,1,1])
I_vv = np.absolute(cov_arr[...,2,2])
if (log_Transform_switch ==True):
arr_lamb1 = np.log10(arr_lamb1)
arr_lamb2 = np.log10(arr_lamb2)
arr_lamb3 = np.log10(arr_lamb3)
co_pol_dif = np.log10(co_pol_dif)
arr_det_cov = np.log10(arr_det_cov)
Rco_X = np.log10(Rco_X)
Ico_X = np.log10(Ico_X)
I_hh = np.log10(I_hh)
I_hv = np.log10(I_hv)
I_vv = np.log10(I_vv)
if(rescale_switch == True):
arr_lamb1=rescale(arr_lamb1)
arr_lamb2=rescale(arr_lamb2)
arr_lamb3=rescale(arr_lamb3)
co_pol_dif = rescale(co_pol_dif)
arr_det_cov = rescale(arr_det_cov)
Rco_X = rescale(Rco_X)
Ico_X = rescale(Ico_X)
I_hh = rescale(I_hh)
I_hv = rescale(I_hv)
I_vv = rescale(I_vv)
#if(log_Transform_switch ==True):
#return 10*np.log10(np.dstack((arr_lamb1, arr_lamb2, arr_lamb3, co_pol_dif, arr_det_cov, Rco_X, Ico_X, I_hh, I_hv, I_vv)))
#else:
return np.dstack((arr_lamb1, arr_lamb2, arr_lamb3, co_pol_dif, arr_det_cov, Rco_X, Ico_X, I_hh, I_hv, I_vv))
def Pauli_RGB_array(window_size, correction, degree):
cov_arr=extract_polarimetric.extract_covariance_arr(window_size, correction, degree)
clip_extremes=True
ShhShh=cov_arr[:,:,0,0]
SvvSvv=cov_arr[:,:,2,2]
ShvShv=cov_arr[:,:,1,1]
ShhSvv_=cov_arr[:,:,0,2]
Shh_Svv=cov_arr[:,:,2,0]
co_pol_sum_mag=ShhShh+SvvSvv+ShhSvv_+Shh_Svv
co_pol_diff_mag=ShhShh+SvvSvv-ShhSvv_-Shh_Svv
#b=plotting.hist_stretch_all(np.real(co_pol_sum_mag)/2, 0, clip_extremes)
#r=plotting.hist_stretch_all(np.real(co_pol_diff_mag)/2, 0, clip_extremes)
#g=plotting.hist_stretch_all(np.real(ShvShv), 0, clip_extremes)
b=np.real(co_pol_sum_mag)/2
r=np.real(co_pol_diff_mag)/2
g=np.real(ShvShv)
img_arr=np.dstack((r,g,b))
#print(np.imag(co_pol_sum_mag))
#print(np.imag(co_pol_diff_mag))
return img_arr
def plot_all_UAVSAR_features(directory_MLC, window_size, correction_switch,
degree):
os.chdir(directory_MLC)
#ax = plt.subplots(3,3)
matplotlib.rcParams.update({'font.size': 3})
cov_arr = extract_polarimetric.extract_covariance_arr(
window_size, correction_switch, degree)
coh_arr = extract_polarimetric.extract_coherency_arr(
window_size, correction_switch, degree)
#===============1=============
plt.subplot(5, 4, 1)
plt.imshow(10 * np.log10(np.absolute(cov_arr[..., 0, 0])), cmap='gray')
plt.colorbar(label='dB')
plt.ylabel('Azimuth')
plt.title('$I_{HH}$')
plt.subplot(5, 4, 2)
plt.imshow(10 * np.log10(np.absolute(cov_arr[..., 1, 1])), cmap='gray')
plt.colorbar(label='dB')
#plt.ylabel('Azimuth')
plt.title('$I_{HV}$')
plt.subplot(5, 4, 3)
plt.imshow(10 * np.log10(np.absolute(cov_arr[..., 2, 2])), cmap='gray')
plt.colorbar(label='dB')
#plt.ylabel('Azimuth')
plt.title('$I_{VV}$')
#===============1=============
arr_co_pol_pow_ratio = extract_polarimetric.co_pol_power_ratio_1(cov_arr)
plt.subplot(5, 4, 4)
plt.imshow(np.absolute(arr_co_pol_pow_ratio), cmap='gray')
plt.colorbar()
#plt.ylabel('Azimuth')
plt.title('$\gamma_{CO}$')
#===============2=============
Rco_X = np.real(cov_arr[..., 0, 2])
Ico_X = np.imag(cov_arr[..., 0, 2])
plt.subplot(5, 4, 5)
plt.imshow(10 * np.log10(Rco_X), cmap='gray')
plt.colorbar(label='dB')
plt.title('$r_{CO}$')
plt.ylabel('Azimuth')
#===============6=============
plt.subplot(5, 4, 6)
plt.imshow(10 * np.log10(abs(Ico_X)), cmap='gray')
plt.colorbar(label='dB')
plt.title('$|i_{CO}|$')
#===============4=============
plt.subplot(5, 4, 7)
slc_dir = '/home/anurag/Documents/MScProject/SAR/OilSpill/North_Sea_UAVSAR/UAV_norway/UA_norway_00709_15092_000_150610_L090_CX_02'
sd_phase_diff = extract_polarimetric.sd_co_pol_phase_diff(slc_dir)
plt.imshow(np.absolute(sd_phase_diff), cmap='gray')
plt.colorbar()
plt.title('$\phi_{CO}$')
#===============5=============
plt.subplot(5, 4, 8)
plt.imshow(10 * np.log10(extract_polarimetric.determinant_cov(cov_arr)),
cmap='gray')
plt.colorbar(label='dB')
plt.title('$det (C_{FP})$')
#plt.ylabel('Azimuth')
#===============6=============
plt.subplot(5, 4, 9)
plt.imshow(10 *
np.log10(np.absolute(cov_arr[..., 0, 0] - cov_arr[..., 2, 2])),
cmap='gray')
plt.colorbar(label='dB')
plt.title('PD')
plt.ylabel('Azimuth')
#===============7=============
plt.subplot(5, 4, 10)
plt.imshow(np.absolute(cov_arr[..., 1, 1]) / np.absolute(
(cov_arr[..., 0, 0] + cov_arr[..., 2, 2])),
cmap='gray')
plt.colorbar()
plt.title('$P_{X}$')
#===============8=============
eigen_full = extract_polarimetric.eigen_raster_full(
window_size, correction_switch, degree)
arr_lamb1 = eigen_full[:, :, 2]
arr_lamb2 = eigen_full[:, :, 1]
arr_lamb3 = eigen_full[:, :, 0]
#===============8=============
plt.subplot(5, 4, 11)
plt.imshow(10 * np.log10(np.absolute(arr_lamb1)), cmap='gray')
plt.colorbar(label='dB')
plt.title('$\lambda_{1}$')
#===============9=============
plt.subplot(5, 4, 12)
plt.imshow(10 * np.log10(np.absolute(arr_lamb2)), cmap='gray')
plt.colorbar(label='dB')
#plt.ylabel('Azimuth')
plt.title('$\lambda_{2}$')
#===============10=============
plt.subplot(5, 4, 13)
plt.imshow(10 * np.log10(np.absolute(arr_lamb3)), cmap='gray')
plt.colorbar(label='dB')
plt.title('$\lambda_{3}$')
plt.ylabel('Azimuth')
#===============11=============
plt.subplot(5, 4, 14)
arr_ent = extract_polarimetric.entropy(eigen_full)
plt.imshow(10 * np.log10(np.absolute(arr_ent)), cmap='gray')
plt.colorbar()
plt.title('$H$')
#===============12=============
plt.subplot(5, 4, 15)
arr_pol_frac = extract_polarimetric.pol_fraction(eigen_full)
plt.imshow(arr_pol_frac, cmap='gray')
plt.colorbar()
plt.title('$PF$')
#===============13=============
plt.subplot(5, 4, 16)
arr_anisotropy = extract_polarimetric.anisotropy(eigen_full)
plt.imshow(arr_anisotropy, cmap='gray')
plt.colorbar()
#plt.xlabel('Range')
#plt.ylabel('Azimuth')
plt.title('$A$')
#================14==============
plt.subplot(5, 4, 17)
co_pol_corr_arr = extract_polarimetric.co_pol_correlation(cov_arr)
plt.imshow(np.absolute(co_pol_corr_arr), cmap='gray')
plt.colorbar()
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('$Co-pol Correlation$')
#================15==============
plt.subplot(5, 4, 18)
arr_conform_coeff = extract_polarimetric.conformity_coeff(cov_arr)
plt.imshow(np.absolute(arr_conform_coeff), cmap='gray')
plt.colorbar()
plt.xlabel('Range')
plt.title('Conformity Coefficient')
#================16==============
plt.subplot(5, 4, 19)
alpha = extract_polarimetric.mean_alpha_angle(eigen_full, window_size,
correction_switch, degree)
plt.imshow(alpha, cmap='gray')
plt.colorbar(label='degrees')
plt.title('Mean alpha angle')
plt.xlabel('Range')
#================16==============
plt.subplot(5, 4, 20)
PH = extract_polarimetric.pedestal_height(eigen_full)
plt.imshow(PH, cmap='gray')
plt.colorbar()
plt.title('PH')
plt.xlabel('Range')
plt.tight_layout()
plt.savefig(
'/home/anurag/Documents/MScProject/Meetings_ITC/Results/Features/Polarimetric_Features/Inc_corr_applied/Det_cov_conj.tiff',
dpi=300,
box_inches='tight',
papertype='a4',
orientation='portrait')
plt.show()
def plot_covariance_matrix_elements(window_size, inci_switch):
arr = extract_polarimetric.extract_covariance_arr(window_size, inci_switch)
fig, ax = plt.subplots(nrows=3, ncols=3)
elements = [
'C22', 'C33', 'C12_real', 'C12_imag', 'C13_real', 'C13_imag',
'C23_real', 'C23_imag'
]
plt.subplot(3, 3, 1)
#plt.plot(t1, f(t1), 'bo', t2, f(t2), 'k')
plt.imshow(np.real(arr[:, :, 0, 0]), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Ihh')
plt.colorbar()
plt.subplot(3, 3, 2)
#plt.plot(t1, f(t1), 'bo', t2, f(t2), 'k')
plt.imshow(np.real(arr[:, :, 0, 1]), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('ShhShv_real')
plt.colorbar()
plt.subplot(3, 3, 3)
#plt.plot(t1, f(t1), 'bo', t2, f(t2), 'k')
plt.imshow(np.real(arr[:, :, 0, 2]), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Shhvv_real')
plt.colorbar()
plt.subplot(3, 3, 4)
#plt.plot(t1, f(t1), 'bo', t2, f(t2), 'k')
plt.imshow(np.imag(arr[:, :, 1, 0]), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Shhvv_imag')
plt.colorbar()
plt.subplot(3, 3, 5)
plt.imshow(np.real(arr[:, :, 1, 1]), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Ihv')
plt.colorbar()
plt.subplot(3, 3, 6)
plt.imshow(np.real(arr[:, :, 1, 2]), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('ShvSvv_real')
plt.colorbar()
plt.subplot(3, 3, 7)
plt.imshow(np.imag(arr[:, :, 2, 0]), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('ShhSvv_imag')
plt.colorbar()
plt.subplot(3, 3, 8)
plt.imshow(np.imag(arr[:, :, 2, 1]), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('ShvSvv_imag')
plt.colorbar()
plt.subplot(3, 3, 9)
plt.imshow(np.real(arr[:, :, 2, 2]), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Ihv')
plt.colorbar()
#plt.suptitle('C3 elements - No averaging applied')
plt.show()
def extract_UAVSAR_T3(directory, window_size, correction_switch, degree):
os.chdir(directory)
T3 = extract_polarimetric.extract_covariance_arr(window_size,
correction_switch, degree)
return T3
def polarimetric_features(window_size, correction_switch):
#fig, ax = plt.subplots(nrows=2, ncols=3)
#arr=extract_polarimetric.eigen_raster_full(window_size)
cov_arr = extract_polarimetric.extract_covariance_arr(
window_size, correction_switch)
#plt.subplot(2,3,1)
plt.imshow(np.absolute(extract_polarimetric.co_pol_power_ratio_1(cov_arr)),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Ivv/Ihh (Co-pol power ratio)')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'Ivv-Ihh_(Co-pol power ratio)' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(2,3,2)
plt.imshow(incidence_angle_corr.hist_stretch(
np.absolute(extract_polarimetric.determinant_cov(cov_arr)), 6),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Det(Cov)')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'Det(Cov)_stretched' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(2,3,2)
plt.imshow(np.absolute(extract_polarimetric.determinant_cov(cov_arr)),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Det(Cov)')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'Det(Cov)' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(2,3,3)
plt.imshow(np.absolute(extract_polarimetric.co_pol_diff(cov_arr)),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Co-pol diff (Ihh-Ivv)')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'Co-pol diff(Ihh-Ivv)' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(2,3,4)
plt.imshow(np.real(extract_polarimetric.co_pol_cross_product(cov_arr)),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Real(ShhSvv) co-pol cross_product')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'Real(ShhSvv) co-pol cross_product' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(2,3,5)
plt.imshow(np.imag(extract_polarimetric.co_pol_cross_product(cov_arr)),
cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Imag(ShhSvv) Co-pol cross_product')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'Imag(ShhSvv) Co-pol cross_product',
dpi=300)
plt.clf()
def cloude_pottier(window_size):
#fig, ax = plt.subplots(nrows=3, ncols=3)
arr = extract_polarimetric.eigen_raster_full(window_size)
cov_arr = extract_polarimetric.extract_covariance_arr(window_size, False)
#plt.subplot(3,3,1)
plt.imshow(arr[:, :, 2], cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('lambda_1')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'lambda_1' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(3,3,2)
plt.imshow(arr[:, :, 1], cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('lambda_2')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'lambda_2' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(3,3,3)
plt.imshow(arr[:, :, 0], cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('lambda_3')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'lambda_3' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(3,3,4)
plt.imshow(extract_polarimetric.entropy(arr), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Entropy')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'Entropy' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(3,3,5)
plt.imshow(extract_polarimetric.anisotropy(arr), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Anisotropy')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'Anisotropy' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(3,3,6)
plt.imshow(extract_polarimetric.pol_fraction(arr), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('Pol_fraction')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'Pol_fraction' + '.tiff',
dpi=300)
plt.clf()
#plt.subplot(3,3,7)
plt.imshow(extract_polarimetric.pedestal_height(arr), cmap='gray')
plt.xlabel('Range')
plt.ylabel('Azimuth')
plt.title('pedestal_height')
plt.colorbar()
plt.savefig('/home/anurag/Documents/MScProject/Meetings_ITC/Results/' +
'pedestal_height' + '.tiff',
dpi=300)
plt.clf()
def MLC_prod(window_size, correction_switch, degree):
return extract_polarimetric.extract_covariance_arr(window_size,
correction_switch,
degree)
correction_switch=True window_size_fea=9 window_size_smoo=35 window_size_boun=3 ''' degree=1 ncomp=2 inci_corr=False window_size_fea=9 window_size_smoo=50 window_size_boun=3 ''' mlc_orig=cov_arr=extract_polarimetric.extract_covariance_arr(1, True, 1) arr_dstack = read_Pol_features(window_size_fea, correction_switch, degree, rescale_switch = True, log_Transform_switch = False) print(arr_dstack.shape) #================get GLCM features================== feature = np.dstack((arr_dstack, glcm)) print (feature.shape) #=========Visualize_histogram============ #visualize_hist(feature) #sys.exit()
if __name__ == '__main__':
os.chdir(
'../North_Sea_UAVSAR/UAV_norway/UA_norway_00709_15092_000_150610_L090_CX_01/MLC_Python_New/norway_00709_15092_000_150610_L090_CX_01_mlc'
)
window_size_cov, correction_switch, degree = 1, False, 1
num_features = 8
num_classes = 2
m = 5
Niter = 10
Looks = 12 * 3
#Looks=1
beta = 1.4 # smoothening parameter
cov_arr = extract_polarimetric.extract_covariance_arr(
window_size_cov, correction_switch, degree)
cov_df = make_cov_arr_df(window_size_cov, correction_switch, degree)
#print(cov_df)
#print(cov_df)
#
TR = classification.get_training_set(cov_df)
TS = classification.get_test_set(cov_df, padding=25)
#print(TR.shape)
Wishart = Wishart_Likelihood(num_classes,
m,
beta,
cov_arr,
window_size_cov,
Niter,
Looks,