def altitude_range(rpc, x, y, w, h, margin_top, margin_bottom):
"""
Computes an altitude range using SRTM data.
Args:
rpc: instance of the rpc_model.RPCModel class
x, y, w, h: four integers defining a rectangular region of interest
(ROI) in the image. (x, y) is the top-left corner, and (w, h) are the
dimensions of the rectangle.
margin_top: margin (in meters) to add to the upper bound of the range
margin_bottom: margin (usually negative) to add to the lower bound of
the range
Returns:
lower and upper bounds on the altitude of the world points that are
imaged by the RPC projection function in the provided ROI. To compute
these bounds, we use SRTM data. The altitudes are computed with respect
to the WGS84 reference ellipsoid.
"""
# TODO: iterate the procedure used here to get a finer estimation of the
# TODO: bounding box on the ellipsoid and thus of the altitude range. For flat
# TODO: regions it will not improve much, but for mountainous regions there is a
# TODO: lot to improve.
# find bounding box on the ellipsoid (in geodesic coordinates)
lon_m, lon_M, lat_m, lat_M = geodesic_bounding_box(rpc, x, y, w, h)
# if bounding box is out of srtm domain, return coarse altitude estimation
if (lat_m < -60 or lat_M > 60):
print "Out of SRTM domain, returning coarse range from rpc"
return altitude_range_coarse(rpc)
# sample the bounding box with regular step of 3 arcseconds (srtm
# resolution)
ellipsoid_points = sample_bounding_box(lon_m, lon_M, lat_m, lat_M)
# compute srtm height on all these points
srtm = common.run_binary_on_list_of_points(ellipsoid_points,
'srtm4',
env_var=('SRTM4_CACHE',
cfg['srtm_dir']))
h = np.ravel(srtm)
# TODO: choose between the two heuristics implemented here
# srtm data may contain 'nan' values (meaning no data is available there).
# These points are most likely water (sea) and thus their height with
# respect to geoid is 0. Thus we replace the nans with 0.
# TODO: this should not be zero, but the geoid/ellipsoid offset
srtm[np.isnan(h)] = 0
# But for safety we prefer to give up the precise
# altitude estimation in these cases and use the coarse one.
# if np.isnan(np.sum(srtm)):
# return altitude_range_coarse(rpc)
# extract extrema (and add a +-100m security margin)
h_m = np.round(h.min()) + margin_bottom
h_M = np.round(h.max()) + margin_top
return h_m, h_M
def altitude_range(rpc, x, y, w, h, margin_top, margin_bottom):
"""
Computes an altitude range using SRTM data.
Args:
rpc: instance of the rpc_model.RPCModel class
x, y, w, h: four integers defining a rectangular region of interest
(ROI) in the image. (x, y) is the top-left corner, and (w, h) are the
dimensions of the rectangle.
margin_top: margin (in meters) to add to the upper bound of the range
margin_bottom: margin (usually negative) to add to the lower bound of
the range
Returns:
lower and upper bounds on the altitude of the world points that are
imaged by the RPC projection function in the provided ROI. To compute
these bounds, we use SRTM data. The altitudes are computed with respect
to the WGS84 reference ellipsoid.
"""
# TODO: iterate the procedure used here to get a finer estimation of the
# TODO: bounding box on the ellipsoid and thus of the altitude range. For flat
# TODO: regions it will not improve much, but for mountainous regions there is a
# TODO: lot to improve.
# find bounding box on the ellipsoid (in geodesic coordinates)
lon_m, lon_M, lat_m, lat_M = geodesic_bounding_box(rpc, x, y, w, h)
# if bounding box is out of srtm domain, return coarse altitude estimation
if (lat_m < -60 or lat_M > 60):
print "Out of SRTM domain, returning coarse range from rpc"
return altitude_range_coarse(rpc)
# sample the bounding box with regular step of 3 arcseconds (srtm
# resolution)
ellipsoid_points = sample_bounding_box(lon_m, lon_M, lat_m, lat_M)
# compute srtm height on all these points
srtm = common.run_binary_on_list_of_points(ellipsoid_points, 'srtm4',
env_var=('SRTM4_CACHE',
cfg['srtm_dir']))
h = np.ravel(srtm)
# TODO: choose between the two heuristics implemented here
# srtm data may contain 'nan' values (meaning no data is available there).
# These points are most likely water (sea) and thus their height with
# respect to geoid is 0. Thus we replace the nans with 0.
# TODO: this should not be zero, but the geoid/ellipsoid offset
srtm[np.isnan(h)] = 0
# But for safety we prefer to give up the precise
# altitude estimation in these cases and use the coarse one.
# if np.isnan(np.sum(srtm)):
# return altitude_range_coarse(rpc)
# extract extrema (and add a +-100m security margin)
h_m = np.round(h.min()) + margin_bottom
h_M = np.round(h.max()) + margin_top
return h_m, h_M
def geodetic_to_geocentric(lat, lon, alt):
"""
Converts WGS84 ellipsoidal coordinates to geocentric cartesian coordinates.
Args:
lat: latitude, in degrees between -90 and 90
lon: longitude, between -180 and 180
alt: altitude, in meters, above the WGS84 reference ellipsoid
Returns:
x, y, z: the cartesian coordinates of the input point, expressed in the
geocentric frame.
The conversion is made by a commandline tool, CartConvert, which is part of
the GeographicLib library:
http://geographiclib.sourceforge.net/html/intro.html
"""
pts = np.vstack([lat, lon, alt]).T
out = common.run_binary_on_list_of_points(pts, 'CartConvert')
return out[:, 0], out[:, 1], out[:, 2]
def geocentric_to_geodetic(x, y, z):
"""
Converts geocentric cartesian coordinates to WGS84 ellipsoidal coordinates.
Args:
x, y, z: the cartesian coordinates of the input point, expressed in the
geocentric frame.
Returns:
lat: latitude, in degrees between -90 and 90
lon: longitude, between -180 and 180
alt: altitude, in meters, above the WGS84 reference ellipsoid
The conversion is made by a commandline tool, CartConvert, which is part of
the GeographicLib library:
http://geographiclib.sourceforge.net/html/intro.html
"""
pts = np.vstack([x, y, z]).T
out = common.run_binary_on_list_of_points(pts, 'CartConvert', '-r')
return out[:, 0], out[:, 1], out[:, 2]
