def check_coverage(img_src, locations, neighborhood = 1, plot_id_name = 'plot_id'):
u"""
Check whether the given locations have valid data values in the specified
raster dataset. Optionally a specified neighborhood can be checked for
valid data values.
"""
# extracting raster values for locations and the given neighborhood
spectra = gdal_utils.extract_spectra(img_src, locations, neighborhood)
coverage = dict()
for spectrum, location in zip(spectra, locations):
# retrieving plot id
plot_id = location['attributes'][plot_id_name]
# if there are no raster values at all or the mean value of the
# neighborhood does not equal 1 then the location has not exclusively
# valid data values
if type(spectrum) is DictionaryType:
if spectrum is None or spectrum['mean'] != 1:
coverage[plot_id] = 0
else:
coverage[plot_id] = 1
else:
if spectrum == 1:
coverage[plot_id] = 1
else:
coverage[plot_id] = 0
return coverage
def extract_spectra(self, description_field='', verbose=True):
u"""
Extract spectra from source imagery and creates well-defined spectrum
objects.
"""
# extracting spectra as simple lists
spectra = gdal_utils.extract_spectra(self.img_src,
self.extract_locations,
neighborhood=self.neighborhood,
verbose=verbose,
bad_bands=self.bad_bands,
scale_factor=self.slope,
nb_type=self.neighborhood_type)
# converting lists to spectrum objects
# defining list of all extracted spectra
# iterating over location/spectrum pairs
for cp, sp in zip(self.extract_locations, spectra):
# retrieving location id
if cp.has_key('attributes'):
additional_attributes = cp['attributes'].keys()
additional_attributes.remove(self.loc_id_field)
loc_id = cp[self.loc_id_field]
# creating new spectrum object using location id and according coordinates
spectrum = Spectrum(loc_id, (cp['x'], cp['y']))
for aa in additional_attributes:
spectrum.set_attribute(aa, cp['attributes'][aa])
spectrum.set_neighborhood(self.neighborhood)
spectrum.set_neighborhood_type(self.neighborhood_type)
spectrum.set_source(self.img_id)
if description_field:
spectrum.set_description(cp[description_field])
# adding values to spectrum
for gb, val in zip(self.good_bands, sp):
spectrum.set_value(gb, val)
# adding values of bad bands to spectrum
for bb in self.bad_bands:
spectrum.set_invalid(bb)
# adding current spectrum to list of all extracted spectra
self.spectra.append(spectrum)
def extract_spectra(self, description_field = '', verbose = True):
u"""
Extract spectra from source imagery and creates well-defined spectrum
objects.
"""
# extracting spectra as simple lists
spectra = gdal_utils.extract_spectra(self.img_src, self.extract_locations,
neighborhood = self.neighborhood,
verbose = verbose, bad_bands = self.bad_bands,
scale_factor = self.slope, nb_type = self.neighborhood_type)
# converting lists to spectrum objects
# defining list of all extracted spectra
# iterating over location/spectrum pairs
for cp, sp in zip(self.extract_locations, spectra):
# retrieving location id
if cp.has_key('attributes'):
additional_attributes = cp['attributes'].keys()
additional_attributes.remove(self.loc_id_field)
loc_id = cp[self.loc_id_field]
# creating new spectrum object using location id and according coordinates
spectrum = Spectrum(loc_id, (cp['x'], cp['y']))
for aa in additional_attributes:
spectrum.set_attribute(aa, cp['attributes'][aa])
spectrum.set_neighborhood(self.neighborhood)
spectrum.set_neighborhood_type(self.neighborhood_type)
spectrum.set_source(self.img_id)
if description_field:
spectrum.set_description(cp[description_field])
# adding values to spectrum
for gb, val in zip(self.good_bands, sp):
spectrum.set_value(gb, val)
# adding values of bad bands to spectrum
for bb in self.bad_bands:
spectrum.set_invalid(bb)
# adding current spectrum to list of all extracted spectra
self.spectra.append(spectrum)
# setting up a list of plots that are linked with the current strip
strip_locations = list()
# setting up the output for current strip id
strip_output = list()
# retrieving plots that are linked with the current strip
for cp in sorted(cached_plots, key=itemgetter('attributes')):
plot_id = cp['attributes']['plot_id']
if plot_id in image_plt_links[strip_id]:
strip_locations.append(cp)
# extracting spectra from the current image strip for all plots that
# are linked with it
# bands that have been regarded bad are ignored
spectra = gdal_utils.extract_spectra(image_data[strip_id],
strip_locations,
verbose=True,
neighborhood=neighborhood,
bad_bands=bad_bands)
# as there are as much plots for the current strip as there are spectra
# we can simultaneously iterate over them
for strip_location, spectrum in zip(strip_locations, spectra):
plot_id = strip_location['attributes']['plot_id']
# checking whether were in some nxn-neighborhood, i.e. retrieved
# values from a group of pixels
if neighborhood > 1:
single_band_output = list()
# retrieving raw values and basic descriptive statistics
for band_sp in spectrum:
raw = "\t".join([str(f) for f in band_sp['raw']])
mean = str(band_sp['mean'])
tgt_file = "".join((re.search("(.+)_\d_", os.path.basename(image_data[strip_id])).group(1), '_spectra.txt'))
# setting up a list of plots that are linked with the current strip
strip_locations = list()
# setting up the output for current strip id
strip_output = list()
# retrieving plots that are linked with the current strip
for cp in sorted(cached_plots, key = itemgetter('attributes')):
plot_id = cp['attributes']['plot_id']
if plot_id in image_plt_links[strip_id]:
strip_locations.append(cp)
# extracting spectra from the current image strip for all plots that
# are linked with it
# bands that have been regarded bad are ignored
spectra = gdal_utils.extract_spectra(image_data[strip_id], strip_locations, verbose = True, neighborhood = neighborhood, bad_bands = bad_bands)
# as there are as much plots for the current strip as there are spectra
# we can simultaneously iterate over them
for strip_location, spectrum in zip(strip_locations, spectra):
plot_id = strip_location['attributes']['plot_id']
# checking whether were in some nxn-neighborhood, i.e. retrieved
# values from a group of pixels
if neighborhood > 1:
single_band_output = list()
# retrieving raw values and basic descriptive statistics
for band_sp in spectrum:
raw = "\t".join([str(f) for f in band_sp['raw']])
mean = str(band_sp['mean'])
std_dev = str(band_sp['std_dev'])
# putting it all together in output text for a single band