def show_hist(source, bins=10, masked=True, title='Histogram'):
"""
Easily display a histogram with matplotlib.
Parameters
----------
bins : int, optional
Compute histogram across N bins.
data : np.array or rasterio.Band or tuple(dataset, bidx)
Input data to display. The first three arrays in multi-dimensional
arrays are plotted as red, green, and blue.
masked : bool, optional
When working with a `rasterio.Band()` object, specifies if the data
should be masked on read.
title : str, optional
Title for the figure.
"""
if plt is None:
raise ImportError("Could not import matplotlib")
if isinstance(source, (tuple, rasterio.Band)):
arr = source[0].read(source[1], masked=masked)
else:
arr = source
# The histogram is computed individually for each 'band' in the array
# so we need the overall min/max to constrain the plot
rng = arr.min(), arr.max()
if len(arr.shape) is 2:
arr = [arr]
colors = ['gold']
else:
colors = ('red', 'green', 'blue', 'violet', 'gold', 'saddlebrown')
# If a rasterio.Band() is given make sure the proper index is displayed
# in the legend.
if isinstance(source, (tuple, rasterio.Band)):
labels = [str(source[1])]
else:
labels = (str(i + 1) for i in range(len(arr)))
# This loop should add a single plot each band in the input array,
# regardless of if the number of bands exceeds the number of colors.
# The colors slicing ensures that the number of iterations always
# matches the number of bands.
# The goal is to provide a curated set of colors for working with
# smaller datasets and let matplotlib define additional colors when
# working with larger datasets.
for bnd, color, label in zip_longest(arr, colors[:len(arr)], labels):
plt.hist(bnd.flatten(),
bins=bins,
alpha=0.5,
color=color,
label=label,
range=rng)
plt.legend(loc="upper right")
plt.title(title, fontweight='bold')
plt.grid(True)
plt.xlabel('DN')
plt.ylabel('Frequency')
fig = plt.gcf()
fig.show()
def stack(ctx, input, bidx, photometric, output, driver):
"""Stack a number of bands from one or more input files into a
multiband dataset.
Input datasets must be of a kind: same data type, dimensions, etc. The
output is cloned from the first input.
By default, rio-stack will take all bands from each input and write them
in same order to the output. Optionally, bands for each input may be
specified using a simple syntax:
--bidx N takes the Nth band from the input (first band is 1).
--bidx M,N,0 takes bands M, N, and O.
--bidx M..O takes bands M-O, inclusive.
--bidx ..N takes all bands up to and including N.
--bidx N.. takes all bands from N to the end.
Examples, using the Rasterio testing dataset, which produce a copy.
rio stack RGB.byte.tif -o stacked.tif
rio stack RGB.byte.tif --bidx 1,2,3 -o stacked.tif
rio stack RGB.byte.tif --bidx 1..3 -o stacked.tif
rio stack RGB.byte.tif --bidx ..2 RGB.byte.tif --bidx 3.. -o stacked.tif
"""
import numpy as np
verbosity = (ctx.obj and ctx.obj.get('verbosity')) or 2
logger = logging.getLogger('rio')
try:
with rasterio.drivers(CPL_DEBUG=verbosity > 2):
output_count = 0
indexes = []
for path, item in zip_longest(input, bidx, fillvalue=None):
with rasterio.open(path) as src:
src_indexes = src.indexes
if item is None:
indexes.append(src_indexes)
output_count += len(src_indexes)
elif '..' in item:
start, stop = map(lambda x: int(x) if x else None,
item.split('..'))
if start is None:
start = 1
indexes.append(src_indexes[slice(start - 1, stop)])
output_count += len(src_indexes[slice(start - 1, stop)])
else:
parts = list(map(int, item.split(',')))
if len(parts) == 1:
indexes.append(parts[0])
output_count += 1
else:
parts = list(parts)
indexes.append(parts)
output_count += len(parts)
with rasterio.open(input[0]) as first:
kwargs = first.meta
kwargs['transform'] = kwargs.pop('affine')
kwargs.update(driver=driver, count=output_count)
if photometric:
kwargs['photometric'] = photometric
with rasterio.open(output, 'w', **kwargs) as dst:
dst_idx = 1
for path, index in zip(input, indexes):
with rasterio.open(path) as src:
if isinstance(index, int):
data = src.read(index)
dst.write(data, dst_idx)
dst_idx += 1
elif isinstance(index, list):
data = src.read(index)
dst.write(data, range(dst_idx,
dst_idx + len(index)))
dst_idx += len(index)
sys.exit(0)
except Exception:
logger.exception("Failed. Exception caught")
sys.exit(1)
def stack(ctx, files, output, driver, bidx, photometric):
"""Stack a number of bands from one or more input files into a
multiband dataset.
Input datasets must be of a kind: same data type, dimensions, etc. The
output is cloned from the first input.
By default, rio-stack will take all bands from each input and write them
in same order to the output. Optionally, bands for each input may be
specified using a simple syntax:
--bidx N takes the Nth band from the input (first band is 1).
--bidx M,N,0 takes bands M, N, and O.
--bidx M..O takes bands M-O, inclusive.
--bidx ..N takes all bands up to and including N.
--bidx N.. takes all bands from N to the end.
Examples, using the Rasterio testing dataset, which produce a copy.
rio stack RGB.byte.tif -o stacked.tif
rio stack RGB.byte.tif --bidx 1,2,3 -o stacked.tif
rio stack RGB.byte.tif --bidx 1..3 -o stacked.tif
rio stack RGB.byte.tif --bidx ..2 RGB.byte.tif --bidx 3.. -o stacked.tif
"""
verbosity = (ctx.obj and ctx.obj.get('verbosity')) or 2
logger = logging.getLogger('rio')
try:
with rasterio.drivers(CPL_DEBUG=verbosity>2):
output, files = resolve_inout(files=files, output=output)
output_count = 0
indexes = []
for path, item in zip_longest(files, bidx, fillvalue=None):
with rasterio.open(path) as src:
src_indexes = src.indexes
if item is None:
indexes.append(src_indexes)
output_count += len(src_indexes)
elif '..' in item:
start, stop = map(
lambda x: int(x) if x else None, item.split('..'))
if start is None:
start = 1
indexes.append(src_indexes[slice(start-1, stop)])
output_count += len(src_indexes[slice(start-1, stop)])
else:
parts = list(map(int, item.split(',')))
if len(parts) == 1:
indexes.append(parts[0])
output_count += 1
else:
parts = list(parts)
indexes.append(parts)
output_count += len(parts)
with rasterio.open(files[0]) as first:
kwargs = first.meta
kwargs['transform'] = kwargs.pop('affine')
kwargs.update(
driver=driver,
count=output_count)
if photometric:
kwargs['photometric'] = photometric
with rasterio.open(output, 'w', **kwargs) as dst:
dst_idx = 1
for path, index in zip(files, indexes):
with rasterio.open(path) as src:
if isinstance(index, int):
data = src.read(index)
dst.write(data, dst_idx)
dst_idx += 1
elif isinstance(index, list):
data = src.read(index)
dst.write(data, range(dst_idx, dst_idx+len(index)))
dst_idx += len(index)
except Exception:
logger.exception("Exception caught during processing")
raise click.Abort()
def show_hist(source, bins=10, masked=True, title='Histogram'):
"""
Easily display a histogram with matplotlib.
Parameters
----------
bins : int, optional
Compute histogram across N bins.
data : np.array or rasterio.Band or tuple(dataset, bidx)
Input data to display. The first three arrays in multi-dimensional
arrays are plotted as red, green, and blue.
masked : bool, optional
When working with a `rasterio.Band()` object, specifies if the data
should be masked on read.
title : str, optional
Title for the figure.
"""
if plt is None:
raise ImportError("Could not import matplotlib")
if isinstance(source, (tuple, rasterio.Band)):
arr = source[0].read(source[1], masked=masked)
else:
arr = source
# The histogram is computed individually for each 'band' in the array
# so we need the overall min/max to constrain the plot
rng = arr.min(), arr.max()
if len(arr.shape) is 2:
arr = [arr]
colors = ['gold']
else:
colors = ('red', 'green', 'blue', 'violet', 'gold', 'saddlebrown')
# If a rasterio.Band() is given make sure the proper index is displayed
# in the legend.
if isinstance(source, (tuple, rasterio.Band)):
labels = [str(source[1])]
else:
labels = (str(i + 1) for i in range(len(arr)))
# This loop should add a single plot each band in the input array,
# regardless of if the number of bands exceeds the number of colors.
# The colors slicing ensures that the number of iterations always
# matches the number of bands.
# The goal is to provide a curated set of colors for working with
# smaller datasets and let matplotlib define additional colors when
# working with larger datasets.
for bnd, color, label in zip_longest(arr, colors[:len(arr)], labels):
plt.hist(
bnd.flatten(),
bins=bins,
alpha=0.5,
color=color,
label=label,
range=rng
)
plt.legend(loc="upper right")
plt.title(title, fontweight='bold')
plt.grid(True)
plt.xlabel('DN')
plt.ylabel('Frequency')
fig = plt.gcf()
fig.show()
