Skip to content

Commit

Permalink
Add support for reading fov file from h5 uhi
Browse files Browse the repository at this point in the history
  • Loading branch information
@havardlovas
havardlovas committed Apr 15, 2024
1 parent 15d4a84 commit dabcbb0
Show file tree
Hide file tree
Showing 2 changed files with 205 additions and 45 deletions.
4 changes: 2 additions & 2 deletions gref4hsi/tests/test_main_uhi.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -88,7 +88,7 @@
# Settings specific to the pre-processing of UHI data. At present they are hardcoded, but they could be integrated
SettingsPreprocess = namedtuple('SettingsPreprocessing', ['dtype_datacube',
'rotation_matrix_hsi_to_body',
'translation_hsi_to_body',
'translation_body_to_hsi',
'rotation_matrix_alt_to_body',
'translation_alt_to_body',
'config_file_name',
Expand All @@ -103,7 +103,7 @@
[1, 0, 0],
[0, 0, -1]]),
# Boolean being expressing whether to rectify only composite (true) or data cube and composite (false). True is fast.
translation_hsi_to_body = np.array([0, 0, 0]),
translation_body_to_hsi = np.array([0, 0, 0]),
rotation_matrix_alt_to_body = np.array([[0, 1, 0],
[1, 0, 0],
[0, 0, -1]]),
Expand Down
246 changes: 203 additions & 43 deletions gref4hsi/utils/uhi_parsing_utils.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -234,49 +234,69 @@ def read_fov_file_to_param(LAB_CAL_DIR, binning_spatial):

return param_dict

def set_camera_model(config, config_file_path, config_uhi, model_type, binning_spatial):
LAB_CAL_DIR = config['General']['lab_cal_dir'] # Where the fov file is

if model_type == 'cal_dir':
def set_camera_model(config, config_file_path, config_uhi, model_type, binning_spatial, fov_arr = None):
"""_summary_
:param config: _description_
:type config: _type_
:param config_file_path: _description_
:type config_file_path: _type_
:param config_uhi: _description_
:type config_uhi: _type_
:param model_type: _description_
:type model_type: _type_
:param binning_spatial: _description_
:type binning_spatial: _type_
:param fov_arr: _description_, defaults to None
:type fov_arr: _type_, optional
"""
if model_type == 'cal_txt':
# Assumes a csv file with column "View_Angle_Deg" holding the view angles
LAB_CAL_DIR = config['General']['lab_cal_dir'] # Where the fov file is
param_dict = read_fov_file_to_param(LAB_CAL_DIR=LAB_CAL_DIR, binning_spatial=binning_spatial)
# User set rotation matrix
R_hsi_body = config_uhi.rotation_matrix_hsi_to_body

r_zyx = RotLib.from_matrix(R_hsi_body).as_euler('ZYX', degrees=False)

# Euler angle representation (any other would do too)
param_dict['rz'] = r_zyx[0]
param_dict['ry'] = r_zyx[1]
param_dict['rx'] = r_zyx[2]

# Vector from origin of HSI to body origin, expressed in body
# User set
t_hsi_body = config_uhi.translation_hsi_to_body
param_dict['tz'] = t_hsi_body[0]
param_dict['ty'] = t_hsi_body[1]
param_dict['tz'] = t_hsi_body[2]

file_name_xml = 'HSI_' + str(binning_spatial) + 'b.xml'
CAMERA_CALIB_XML_DIR = config['Absolute Paths']['calib_folder']
xml_cal_write_path = CAMERA_CALIB_XML_DIR + file_name_xml
elif model_type == 'embedded':
# Use supplied fov_array
param_dict = Specim.fov_2_param(fov = fov_arr)
else:
pass

CalibHSI(file_name_cal_xml= xml_cal_write_path,
mode = 'w',
param_dict = param_dict)


# Set value in config file and update:
config.set('Relative Paths', 'hsi_calib_path', value = 'Input/Calib/' + file_name_xml)

with open(config_file_path, 'w') as configfile:
config.write(configfile)
# User set rotation matrix
R_hsi_body = config_uhi.rotation_matrix_hsi_to_body

r_zyx = RotLib.from_matrix(R_hsi_body).as_euler('ZYX', degrees=False)

elif model_type == 'embedded':
print('Should make support for this')
pass
else:
pass
# Euler angle representation (any other would do too)
param_dict['rz'] = r_zyx[0]
param_dict['ry'] = r_zyx[1]
param_dict['rx'] = r_zyx[2]

# Vector from origin of body to HSI
t_hsi_body = config_uhi.translation_body_to_hsi
param_dict['tz'] = t_hsi_body[0]
param_dict['ty'] = t_hsi_body[1]
param_dict['tz'] = t_hsi_body[2]

# Define where to write calibrated data
file_name_xml = 'HSI_' + str(binning_spatial) + 'b.xml'
CAMERA_CALIB_XML_DIR = config['Absolute Paths']['calib_folder']
xml_cal_write_path = CAMERA_CALIB_XML_DIR + file_name_xml


CalibHSI(file_name_cal_xml= xml_cal_write_path,
mode = 'w',
param_dict = param_dict)




# Set value in config file and update:
config.set('Relative Paths', 'hsi_calib_path', value = 'Input/Calib/' + file_name_xml)

with open(config_file_path, 'w') as configfile:
config.write(configfile)



Expand Down Expand Up @@ -471,8 +491,26 @@ def altimeter_data_to_point_cloud(nav, config_uhi, lat0, lon0, h0, true_time_hsi

return points_altimeter_transect

def write_point_cloud_to_dem(points, config, resolution_dem, lon0, lat0, h0, method='nearest', smooth_DEM=True):

def point_cloud_to_dem(points, config, resolution_dem, lon0, lat0, h0, method='nearest', smooth_DEM=True):
"""Converts the sparse point cloud of altimeter data into a digital elevation model.
:param points: _description_
:type points: _type_
:param config: _description_
:type config: _type_
:param resolution_dem: _description_
:type resolution_dem: _type_
:param lon0: _description_
:type lon0: _type_
:param lat0: _description_
:type lat0: _type_
:param h0: _description_
:type h0: _type_
:param method: _description_, defaults to 'nearest'
:type method: str, optional
:param smooth_DEM: _description_, defaults to True
:type smooth_DEM: bool, optional
"""
# The name of the digital elevation model
output_dem_path = config['Absolute Paths']['dem_path']

Expand Down Expand Up @@ -510,7 +548,7 @@ def write_point_cloud_to_dem(points, config, resolution_dem, lon0, lat0, h0, met
# Transform extent
(x, y, z) = transformer.transform(xx=lat_ext, yy=lon_ext, zz=h_ext)

if crs_dem.is_geocentric:
if crs_dem.is_geographic:

res_x = resolution_dem*(x.max() - x.min())/(x_max - x_min)
res_y = resolution_dem*(y.max() - y.min())/(y_max - y_min)
Expand Down Expand Up @@ -573,12 +611,13 @@ def write_point_cloud_to_dem(points, config, resolution_dem, lon0, lat0, h0, met

if smooth_DEM == True:
from scipy import ndimage
# Smooths by 4 pixel units
sigma_x = 4
sigma_y = 4
grid_z = ndimage.gaussian_filter(grid_z, sigma=(sigma_y, sigma_x))



# Writes to a file
with rasterio.open(output_dem_path, 'w', driver='GTiff', height=grid_z.shape[0],
width=grid_z.shape[1], count=1, dtype='float32', crs='EPSG:' + str(dem_epsg),
transform=transform) as dst:
Expand Down Expand Up @@ -649,8 +688,9 @@ def uhi_beast(config, config_uhi):
set_camera_model(config=config,
config_file_path = config_file_path,
config_uhi=config_uhi,
model_type = 'cal_dir',
binning_spatial = binning_spatial)
model_type = 'cal_txt',
binning_spatial = binning_spatial,
hyp_obj=hyp)

true_time_hsi = hyp.hsi_frames_timestamp - time_offset

Expand Down Expand Up @@ -703,15 +743,135 @@ def uhi_beast(config, config_uhi):



write_point_cloud_to_dem(point_cloud_altimeter_total,
point_cloud_to_dem(point_cloud_altimeter_total,
config,
resolution_dem = config_uhi.resolution_dem,
lon0=lon0,
lat0=lat0,
h0=alt0)




def uhi_dbe(config, config_uhi):
"""Preparing data for the UHI-2x blueye edition (almost copy of beast)
:param config: _description_
:type config: _type_
:param config_uhi: _description_
:type config_uhi: _type_
"""
MISSION_PATH = config['General']['mission_dir'] # Where h5 data is


# For writing
config_file_path = os.path.join(MISSION_PATH, 'configuration.ini')

SPATIAL_PIXELS = 1936 # Same for almost all UHI


INPUT_DIR = MISSION_PATH + 'Input/'

h5_folder = config['Absolute Paths']['h5_folder']
H5_PATTERN = '*.h5'

MAT_DIR = INPUT_DIR
MAT_PATTERN = '*.mat'

# Locate mat file with nav data
search_path_mat = os.path.normpath(os.path.join(MAT_DIR, MAT_PATTERN))
MAT_FILE_PATHS = glob.glob(search_path_mat)
MAT_PATH = MAT_FILE_PATHS[0] # Assuming there is only one


# TODO: replace with how you read in data from DVL/H5 folders
nav = read_nav_from_mat(mat_filename=MAT_PATH)



# Search the h5 folder (these are the files to iterate)
search_path_h5 = os.path.normpath(os.path.join(h5_folder, H5_PATTERN))
H5_FILE_PATHS = glob.glob(search_path_h5)

number_of_h5_files = len(H5_FILE_PATHS)

h5_dict_read = {'radiance_cube': config['HDF.hyperspectral']['datacube'],
'hsi_frames_timestamp': config['HDF.hyperspectral']['timestamp'],
'fov': config['HDF.calibration']['fov'],
'wavelengths' : config['HDF.calibration']['band2wavelength'],
#'rgb_frames' : config['HDF.rgb']['rgb_frames'],
#'rgb_timestamp' : config['HDF.rgb']['rgb_timestamp']
}

time_offset = config_uhi.time_offset_sec
lon0, lat0, alt0 = config_uhi.lon_lat_alt_origin

# from gref4hsi.utils.photogrammetry_utils import Photogrammetry
# agisoft_object = Photogrammetry(project_folder = MISSION_PATH, software_type='agisoft')
# TODO:: Add support for concurrent camera given that images are contained in a h5 folder.
# Should interface towards ODM as well

for h5_index in range(number_of_h5_files):
H5_FILE_PATH = H5_FILE_PATHS[h5_index]

# Read the specified entries
hyp = HyperspectralLite(h5_filename=H5_FILE_PATH, h5_tree_dict=h5_dict_read)

if h5_index == 0:
# Camera model is set once, assuming same spatial binning throughout
binning_spatial = int(np.round(SPATIAL_PIXELS/hyp.radiance_cube.shape[1]))

# Extract from the hyperspectral object
fov_arr = hyp.fov

# Sets camera model with user specified boresight ...
set_camera_model(config=config,
config_file_path = config_file_path,
config_uhi=config_uhi,
model_type = 'embedded',
binning_spatial = binning_spatial,
fov_arr=fov_arr)

true_time_hsi = hyp.hsi_frames_timestamp - time_offset

# Interpolate nav to roll time, IMU time (considered nav timestamp)

## write nav data to h5 file
write_nav_data_to_h5(nav, time_offset, config, H5_FILE_PATH)

# Build a point cloud
point_cloud_altimeter = altimeter_data_to_point_cloud(nav = nav,
config_uhi=config_uhi,
true_time_hsi = true_time_hsi,
lon0=lon0,
lat0=lat0,
h0=alt0)
if h5_index == 0:
point_cloud_altimeter_total = point_cloud_altimeter
else:
point_cloud_altimeter_total = np.append(point_cloud_altimeter_total, point_cloud_altimeter, axis = 0)



print(H5_FILE_PATH)

# Use the total point cloud to make a DEM


if config_uhi.agisoft_process:
agisoft_object.load_photos_and_reference()





point_cloud_to_dem(point_cloud_altimeter_total,
config,
resolution_dem = config_uhi.resolution_dem,
lon0=lon0,
lat0=lat0,
h0=alt0)




0 comments on commit dabcbb0

Please sign in to comment.