def test_mask_exact2(test_fixture2):
with rio.open(test_fixture2) as src:
img = src.read()
ndv = (18, 51, 62)
assert np.max(mask_exact(img, ndv)) <= np.iinfo(img.dtype).max
assert mask_exact(img, ndv).dtype == img.dtype
assert np.array_equal(
np.invert(np.all(np.rollaxis(img, 0, 3) == ndv, axis=2)).astype(
img.dtype) * np.iinfo(img.dtype).max, mask_exact(img, ndv))
def test_mask_exact3(arr):
for i in range(3):
arr[i][-1][-1] = 0
ndv = (0, 0, 0)
assert np.max(mask_exact(arr, ndv)) <= np.iinfo(arr.dtype).max
assert mask_exact(arr, ndv).dtype == arr.dtype
assert np.array_equal(
np.invert(np.all(np.rollaxis(arr, 0, 3) == ndv, axis=2)).astype(
arr.dtype) * np.iinfo(arr.dtype).max, mask_exact(arr, ndv))
def count_ndv_regions(img, ndv):
'''Discover unique labels to count ndv regions.
Parameters
----------
ndv: list
a list of floats whose length = band count
img: ndarray
(depth x rows x cols) array
Returns
-------
n_labels: int
an integer equal to the number of connected regions
'''
np.set_printoptions(threshold=np.nan)
img = mask_exact(img, ndv)
_, n_labels = measure.label(img,
background=255,
neighbors=4,
return_num=True)
return n_labels
def test_mask_exact4():
arr2 = np.random.randint(0, 255, (3, 10, 15))
for i in range(3):
arr2[i][-1][-1] = 255
arr2[i][1][1] = 255
arr2[i][2][2] = 255
ndv = (255, 255, 255)
assert np.array_equal(np.any(np.rollaxis(arr2, 0, 3) != ndv, axis=2),
mask_exact(arr2, ndv) / np.iinfo(arr2.dtype).max)
def _alpha_worker(open_file, window, ij, g_args):
"""rio mucho worker for alpha. It reads input
files and perform alpha calculations on each window.
Parameters
------------
open_files: list of rasterio open files
window: tuples
A window is a view onto a rectangular subset of a
raster dataset and is described in rasterio
by a pair of range tuples:
((row_start, row_stop), (col_start, col_stop))
g_args: dictionary
Returns
---------
rgba: ndarray
ndarray with original RGB bands of shape (3, rows, cols)
and a mask of shape (rows, cols) where
opaque == 0 and transparent == max of dtype
"""
src = open_file[0]
arr = src.read(window=window)
# Determine Alpha Band
if g_args['ndv']:
# User-supplied nodata value
alpha = mask_exact(arr, g_args['ndv'])
else:
# Let rasterio decide
alpha = src.dataset_mask(window=window)
# Replace or Add alpha band to input data
if arr.shape[0] == 4:
# replace band 4 with new alpha mask
# (likely the same but let's not make that assumption)
rgba = arr.copy()
rgba[3] = alpha
elif arr.shape[0] == 3:
# stack the alpha mask to add band 4
rgba = np.append(arr, alpha[np.newaxis, :, :], axis=0)
else:
raise ValueError("Array must have 3 or 4 bands (RGB or RGBA)")
return rgba
def count_ndv_regions(img, ndv):
"""Discover unique labels to count ndv regions.
Parameters
----------
ndv: list
a list of floats whose length = band count
img: ndarray
(depth x rows x cols) array
Returns
-------
int
The number of connected regions
"""
img = mask_exact(img, ndv)
return len(list(shapes(img, mask=(img != 255))))