def test_param_extraction(sample_subdirectory,sample_image_path,regex_images):
""" Plot the output of extract_select_parameters.extract_fmin_and_max_depth
:param sample_subdirectory: sample subdir to test coord_extraction and map_coordinates_positions_to_values
:type sample_subdirectory: str
:param sample_image_path: path of the sample image
:type sample_image_path: str
:param regex_images: regular expression to extract images ex:'E:/master/R*/[0-9]*/'
:type regex_images: str
"""
# Segment images in the subdirectory
df_img,_,_ =segment_images(sample_subdirectory, regex_images)
# Get stack
stack = all_stack(df_img)
col_peaks,row_peaks,mapping_Hz, mapping_km = get_grid_mappings(stack)
# Get random ionogram
df_sample = df_img[df_img['file_name'] == sample_image_path]
row = df_sample.iloc[0,:]
ionogram = row['ionogram']
# Get (x,y) coordinates of trace
raw_coord, window_coord = extract_coord(ionogram ,col_peaks,row_peaks)
# (Hz, km) coordinates
arr_adjusted_coord = map_coordinates_positions_to_values(window_coord,col_peaks,row_peaks,mapping_Hz,mapping_km)
# Extract parameters
fmin, depth_max = extract_fmin_and_max_depth(arr_adjusted_coord)
# For plotting purposes
# remove outliers ie coordinates less coordinates corresponding to 0.5 Hz or more than corresponding to 11.5 Hz
col_peaks = np.array(list(mapping_Hz.values())) # use the modified col_peaks ie the one with exactly 13 values
mask = np.logical_or(window_coord[:,0] < col_peaks.min(), window_coord[:,0] > col_peaks.max())
window_coord_adjusted = window_coord[~mask,:]
fmin_coord, depth_max_coord = extract_fmin_and_max_depth(window_coord_adjusted,if_raw=True)
# Plots
fig,axes = plt.subplots(nrows=1,ncols=2)
ax= axes.ravel()
fig.suptitle(row['file_name'])
ax[0].set_title("Original ionogram " )
ax[0].imshow(ionogram,'gray')
colored_iono = cv2.cvtColor(ionogram ,cv2.COLOR_GRAY2RGB)
h,w = ionogram.shape
cv2.line(colored_iono,(int(fmin_coord),0),(int(fmin_coord),h),[0,0,255],5)
cv2.line(colored_iono,(0,int(depth_max_coord)),(w,int(depth_max_coord)),[0,0,255],5)
ax[1].imshow(colored_iono)
ax[1].set_title('fmin='+ str(fmin) + 'max_depth='+ str(depth_max))
plt.axis('off')
def test_coord_extraction(sample_subdirectory,sample_image_path,regex_images,to_overlay):
""" Plot the output of extract_all_coordinates_ionogram_trace.extract_coord and extract_all_coordinates_ionogram_trace.map_coordinates_positions_to_values on a random ionogram
:param sample_subdirectory: sample subdir to test coord_extraction and map_coordinates_positions_to_values
:type sample_subdirectory: str
:param sample_image_path: path of the sample image
:type sample_image_path: str
:param regex_images: regular expression to extract images ex:'E:/master/R*/[0-9]*/'
:type regex_images: str
:param to_overlay: whether to plot on raw ionogram
:type to_overlay: bool
"""
# Segment images in the subdirectory
df_img,_,_ =segment_images(sample_subdirectory, regex_images)
# Get stack
stack = all_stack(df_img)
col_peaks,row_peaks,mapping_Hz, mapping_km = get_grid_mappings(stack)
# Get data from select raw image
df_sample = df_img[df_img['file_name'] == sample_image_path]
row = df_sample.iloc[0,:]
ionogram = row['ionogram']
# Get coordinates of trace
raw_coord, window_coord = extract_coord(ionogram ,col_peaks,row_peaks)
# (Hz, km) coordinates
arr_adjusted_coord = map_coordinates_positions_to_values(window_coord,col_peaks,row_peaks,mapping_Hz,mapping_km)
# Plots
fig,axes = plt.subplots(nrows=2,ncols=2)
ax= axes.ravel()
fig.suptitle(row['file_name'])
ax[0].set_title("Original ionogram " )
ax[0].imshow(ionogram,'gray')
if to_overlay:
ax[1].set_title("Extracted trace " )
ax[1].imshow(ionogram,'gray')
ax[1].scatter(list(zip(*raw_coord))[0],list(zip(*raw_coord))[1],s=1)
ax[2].set_title("Windowed extracted trace " )
ax[2].imshow(ionogram,'gray')
ax[2].scatter(list(zip(*window_coord))[0],list(zip(*window_coord))[1],s=1)
for i in range(0,3):
ax[i].axis('off')
else:
ax[1].set_title("Extracted trace " )
ax[1].scatter(list(zip(*raw_coord))[0],list(zip(*raw_coord))[1],s=1)
ax[1].invert_yaxis()
ax[2].set_title("Windowed extracted trace " )
ax[2].scatter(list(zip(*window_coord))[0],list(zip(*window_coord))[1],s=1)
ax[2].invert_yaxis()
for i in range(0,3):
ax[i].axis('off')
ax[3].set_title("Labeled extracted trace " )
ax[3].scatter(list(zip(*arr_adjusted_coord))[0],list(zip(*arr_adjusted_coord))[1],s=1)
ax[3].invert_yaxis()
ax[3].set_xlabel('Frequency (Hz)')
ax[3].set_ylabel('Depth (km)')
def plot_curve_extraction(regex_subdir,regex_images,to_overlay):
"""Plot the output of subtraces_extraction on a random ionogram
:param regex_subdir: regular expression to extract subdirectory ex: 'E:/master/R*/[0-9]*/'
:type regex_img: str
:param regex_img: regular expression to extract images ex: '*.png'
:type regex_img: str
:param to_overlay: whether to plot on raw ionogram
:type to_overlay: bool
"""
# All the subdirectory i.e. R014207948/1743-9/
list_all_subdir = glob.glob(regex_subdir)
# Randomly pick a subdirectory
sample_subdir = list_all_subdir[random.randint(0,len(list_all_subdir) - 1)]
# Segment images in the subdirectory
df_img,_,_ =segment_images(sample_subdir, regex_images)
# Get stack
stack = all_stack(df_img)
col_peaks,row_peaks,mapping_Hz, mapping_km = get_grid_mappings(stack)
# Get random ionogram
df_sample = df_img.sample(n=1)
ionogram = df_sample['ionogram'].iloc[0]
# Get coordinates of trace
raw_coord, window_coord = extract_coord(ionogram ,col_peaks,row_peaks)
# Plot clusterings
fig,axes = plt.subplots(nrows=2,ncols=2)
fig.suptitle(df_sample['file_name'].iloc[0])
ax= axes.ravel()
ax[0].set_title("Original ionogram " )
ax[0].imshow(ionogram,'gray')
ax[3].set_title("Clustered extracted trace" )
if to_overlay:
ax[1].set_title("Extracted trace " )
ax[1].imshow(ionogram,'gray')
ax[1].scatter(list(zip(*raw_coord))[0],list(zip(*raw_coord))[1],s=1)
ax[2].set_title("Windowed extracted trace " )
ax[2].imshow(ionogram,'gray')
ax[2].scatter(list(zip(*window_coord))[0],list(zip(*window_coord))[1],s=1)
ax[3].imshow(ionogram,'gray')
df_arr_raw_coord = subtraces_extraction(window_coord )
ax[3].scatter(list(zip(*window_coord))[0],list(zip(*window_coord))[1],s=1, label='raw')
ax[3].scatter(df_arr_raw_coord['Hz'].tolist(),df_arr_raw_coord['Median'].tolist(),s=5, label='median')
ax[3].scatter(df_arr_raw_coord ['Hz'].tolist(),df_arr_raw_coord['Cluster1_median'].tolist(),s=5,label='median1')
ax[3].scatter(df_arr_raw_coord ['Hz'].tolist(),df_arr_raw_coord['Cluster2_median'].tolist(),s=5,label='median2')
for i in range(0,4):
ax[i].axis('off')
else:
ax[1].set_title("Windowed extracted trace " )
ax[1].scatter(list(zip(*window_coord))[0],list(zip(*window_coord))[1],s=1)
ax[1].invert_yaxis()
arr_adjusted_coord = map_coordinates_positions_to_values(window_coord,col_peaks,row_peaks,mapping_Hz,mapping_km)
ax[2].set_title("Labeled extracted trace " )
ax[2].scatter(list(zip(*arr_adjusted_coord))[0],list(zip(*arr_adjusted_coord))[1],s=1)
df_arr_adjusted_coord = subtraces_extraction(arr_adjusted_coord)
ax[3].scatter(list(zip(*arr_adjusted_coord ))[0],list(zip(*arr_adjusted_coord ))[1],s=1, label='raw')
ax[3].scatter(df_arr_adjusted_coord['Hz'].tolist(),df_arr_adjusted_coord['Median'].tolist(),s=5, label='median')
ax[3].scatter(df_arr_adjusted_coord ['Hz'].tolist(),df_arr_adjusted_coord['Cluster1_median'].tolist(),s=5,label='median1')
ax[3].scatter(df_arr_adjusted_coord ['Hz'].tolist(),df_arr_adjusted_coord['Cluster2_median'].tolist(),s=5,label='median2')
for i in range(0,2):
ax[i].axis('off')
for i in range(2,4):
ax[i].invert_yaxis()
ax[i].set_xlabel('Frequency (Hz)')
ax[i].set_ylabel('Depth (km)')
plt.legend(loc='best')
def test_extract_coord_subdir_and_param(sample_subdirectory,sample_image_path,regex_images,to_overlay):
""" Plot the output of extract_all_coordinates_ionogram_trace.extract_coord_subdir_and_param
:param sample_subdirectory: sample subdir to test coord_extraction and map_coordinates_positions_to_values
:type sample_subdirectory: str
:param sample_image_path: path of the sample image
:type sample_image_path: str
:param regex_images: regular expression to extract images ex:'E:/master/R*/[0-9]*/'
:type regex_images: str
:param to_overlay: whether to plot on raw ionogram
:type to_overlay: bool
"""
# Segment images in the subdirectory
df_img,_,_ =segment_images(sample_subdirectory, regex_images)
# Get stack
stack = all_stack(df_img)
col_peaks,row_peaks,mapping_Hz, mapping_km = get_grid_mappings(stack)
df_img, df_loss_coord = extract_coord_subdir_and_param(df_img,sample_subdirectory,col_peaks,row_peaks,mapping_Hz,mapping_km)
# Visualize a random row of the dataframe output of df_img
df_sample = df_img[df_img['file_name'] == sample_image_path]
row = df_sample.iloc[0,:]
fig,axes = plt.subplots(nrows=2,ncols=2)
ax=axes.ravel()
ax[0].set_title("Original ionogram " )
ax[0].imshow(row['ionogram'],'gray')
if to_overlay:
ax[1].set_title("Windowed extracted trace " )
ax[1].imshow(row['ionogram'],'gray')
ax[1].scatter(list(zip(*row['window_coord']))[0],list(zip(*row['window_coord']))[1],s=1)
else:
ax[1].set_title("Windowed extracted trace " )
ax[1].scatter(list(zip(*row['window_coord']))[0],list(zip(*row['window_coord']))[1],s=1)
ax[1].invert_yaxis()
colored_iono = cv2.cvtColor(row['ionogram'] ,cv2.COLOR_GRAY2RGB)
h,w = row['ionogram'].shape
col_peaks = np.array(list(mapping_Hz.values())) # use the modified col_peaks ie the one with exactly 13 values
mask = np.logical_or(row['window_coord'][:,0] < col_peaks.min(), row['window_coord'][:,0] > col_peaks.max())
window_coord_adjusted = row['window_coord'][~mask,:]
fmin_coord, depth_max_coord = extract_fmin_and_max_depth(window_coord_adjusted,if_raw=True)
cv2.line(colored_iono,(int(fmin_coord),0),(int(fmin_coord),h),[0,0,255],5)
cv2.line(colored_iono,(0,int(depth_max_coord)),(w,int(depth_max_coord)),[0,0,255],5)
ax[2].imshow(colored_iono)
ax[2].set_title('fmin='+ str(row['fmin']) + 'max_depth='+ str(row['max_depth']))
for i in range(0,3):
ax[i].axis('off')
ax[3].set_title("Labeled extracted trace " )
ax[3].scatter(list(zip(*row['mapped_coord']))[0],list(zip(*row['mapped_coord']))[1],s=1)
ax[3].invert_yaxis()
ax[3].set_xlabel('Frequency (Hz)')
ax[3].set_ylabel('Depth (km)')
if __name__ == '__main__':
regex_subdir = 'E:/master/R*/[0-9]*/'
regex_images = '*.png'
# All the subdirectory i.e. R014207948/1743-9/
list_all_subdir = glob.glob(regex_subdir)
# Randomly pick a subdirectory
sample_subdir = list_all_subdir[random.randint(0,
len(list_all_subdir) - 1)]
# Segment images in the subdirectory
df_img, _, _ = segment_images(sample_subdir, regex_images)
# Get stack
stack = all_stack(df_img)
col_peaks, row_peaks, mapping_Hz, mapping_km = get_grid_mappings(stack)
# Get random ionogram
df_sample = df_img.sample(n=1)
ionogram = df_sample['ionogram'].iloc[0]
# Get coordinates of trace
raw_coord, window_coord = extract_coord(ionogram, col_peaks, row_peaks)
# Get (Hz, km) coordinates from ionogram
arr_adjusted_coord = map_coordinates_positions_to_values(
window_coord, col_peaks, row_peaks, mapping_Hz, mapping_km)
# Plot curve fittings
plot_curve_fitting(ionogram, arr_adjusted_coord)
plot_curve_fitting(ionogram, arr_adjusted_coord, log_x=True)