def fill_arr(a, win=(3, 3)):
"""try filling an array"""
fd = np.array([[32, 64, 128], [16, 0, 1], [8, 4, 2]]) # flow direction
# if (zone < a.min()) or (zone > a.max()) or (zone is None):
# print("\nYou need a zone that is within the range of values.")
# return a, None
if win[0] == 3:
pr = 1
else:
pr = 0
ap = np.pad(a, pad_width=(1, pr), mode="constant", constant_values=(0, 0))
if win == (2, 2):
a_c = ap[1:, 1:] # for 2x2 even
w, h = win
elif win == (3, 3):
w, h = win
a_c = ap[1:-1, 1:-1] # for 3x3 odd
a_s = stride(a_c, win=win) # stride the array
r, c = a_s.shape[:2]
out = []
x = a_s.shape[0]
y = a_s.shape[1]
for i in range(x):
for j in range(y):
# do stuff
sub = a_s[i, j].ravel()
edges = np.asarray([sub[:4], sub[5:]]).ravel()
e_min = edges[np.argmin(edges)]
if sub[4] < e_min:
out.append(e_min)
else:
out.append(sub[4])
out = np.asarray(out).reshape(r, c)
return out # , a_s, ap, a_c
def fill_arr(a, win=(3, 3)):
"""try filling an array
as in fill, sinks
"""
#fd = np.array([[32, 64, 128], [16, 0, 1], [8, 4, 2]]) # flow direction
# if (zone < a.min()) or (zone > a.max()) or (zone is None):
# print("\nYou need a zone that is within the range of values.")
# return a, None
if win[0] == 3:
pr = 1
else:
pr = 0
ap = np.pad(a, pad_width=(1, pr), mode="constant", constant_values=(0, 0))
if win == (2, 2):
a_c = ap[1:, 1:] # for 2x2 even
elif win == (3, 3):
a_c = ap[1:-1, 1:-1] # for 3x3 odd
a_s = stride(a_c, win=win) # stride the array
r, c = a_s.shape[:2]
out = []
x = a_s.shape[0]
y = a_s.shape[1]
for i in range(x):
for j in range(y):
# do stuff
sub = a_s[i, j].ravel()
edges = np.asarray([sub[:4], sub[5:]]).ravel()
e_min = edges[np.argmax(edges)] # argmax or argmin???
if sub[4] < e_min:
out.append(e_min)
else:
out.append(sub[4])
out = np.asarray(out).reshape(r, c)
return out # , a_s, ap, a_c
def _slope_aspect_demo_(cell_size=2):
"""Demo of the data set below"""
a = np.array([[0, 1, 2, 3, 3, 3, 2, 1, 0], [1, 2, 3, 4, 4, 4, 3, 2, 1],
[2, 3, 4, 5, 5, 5, 4, 3, 2], [3, 4, 5, 5, 5, 5, 5, 4, 3],
[3, 4, 5, 5, 5, 5, 5, 4, 3], [3, 4, 5, 5, 5, 5, 5, 4, 3],
[2, 3, 4, 5, 5, 5, 4, 3, 2], [1, 2, 3, 4, 4, 4, 3, 2, 1],
[0, 1, 2, 3, 3, 3, 2, 1, 0]])
r, c = a.shape
data = stride(a, win=(3, 3), stepby=(1, 1)) # produce strided array
slope = slope_a(a, cell_size=cell_size)
slope = np.array(slope)
aspect = [
aspect_a(data[i][j]) for i in range(data.shape[0])
for j in range(data.shape[1])
]
aspect = np.array(aspect).reshape((r - 2, c - 2))
frmt = """
:---- Slope, Aspect Demo ------------------------------------------------
:Sample DEM with a cell size of {} units.
{}\n
Slope (degrees) ...
{}\n
Aspect (degrees) ...
{}
"""
args = [cell_size]
pre = ' ..'
args.extend([indent(str(i), pre) for i in [a, slope, aspect]])
print(dedent(frmt).format(*args))
return a, slope, aspect
def _slope_aspect_demo_(cell_size=2):
"""Demo of the data set below"""
a = np.array([[0, 1, 2, 3, 3, 3, 2, 1, 0],
[1, 2, 3, 4, 4, 4, 3, 2, 1],
[2, 3, 4, 5, 5, 5, 4, 3, 2],
[3, 4, 5, 5, 5, 5, 5, 4, 3],
[3, 4, 5, 5, 5, 5, 5, 4, 3],
[3, 4, 5, 5, 5, 5, 5, 4, 3],
[2, 3, 4, 5, 5, 5, 4, 3, 2],
[1, 2, 3, 4, 4, 4, 3, 2, 1],
[0, 1, 2, 3, 3, 3, 2, 1, 0]])
r, c = a.shape
data = stride(a, win=(3, 3), stepby=(1, 1)) # produce strided array
slope = slope_a(a, cell_size=cell_size)
slope = np.array(slope)
aspect = [aspect_a(data[i][j])
for i in range(data.shape[0])
for j in range(data.shape[1])]
aspect = np.array(aspect).reshape((r-2, c-2))
frmt = """
:---- Slope, Aspect Demo ------------------------------------------------
:Sample DEM with a cell size of {} units.
{}\n
Slope (degrees) ...
{}\n
Aspect (degrees) ...
{}
"""
args = [cell_size]
pre = ' ..'
args.extend([indent(str(i), pre) for i in [a, slope, aspect]])
print(dedent(frmt).format(*args))
return a, slope, aspect
def aspect_a(a, cell_size=1, flat=0.1, degrees=True, keepdims=False):
"""Return the aspect of a slope in degrees from North.
:Requires:
:--------
: a - an input 2d array. X and Y represent coordinates of the Z values
: cell_size - needed to proper flat calculation
: flat - degree value, e.g. flat surface <= 0.05 deg
: 0.05 deg => 8.7e-04 rad 0.10 deg => 1.7e-02 rad
:
"""
if not isinstance(flat, (int, float)):
flat = 0.1
a_s = stride(a, win=(3, 3), stepby=(1, 1))
if a_s.ndim < 4:
new_shape = (1, ) * (4 - len(a_s.shape)) + a_s.shape
a_s = a_s.reshape(new_shape)
f_dxyz = np.array([[1, 2, 1], [2, 0, 2], [1, 2, 1]], dtype="float64")
a_s = a_s * f_dxyz
#
dz_dx, dz_dy = filter_a(a_s, a_filter=f_dxyz, cell_size=1)
#
asp = np.arctan2(dz_dy, -dz_dx) # relative to East
# get the slope
s = np.sqrt((dz_dx * cell_size)**2 + (dz_dy * cell_size)**2)
asp = np.rad2deg(asp)
asp = np.mod((450.0 - asp), 360.) # simplest way to get azimuth
asp = np.where(s <= flat, -1, asp)
if not keepdims:
asp = np.squeeze(asp)
if not degrees:
asp = np.deg2rad(asp)
return asp
def slope_a(a, cell_size=1, kern=None, degrees=True, verb=False, keep=False):
"""Return slope in degrees for an input array using 3rd order
finite difference method for a 3x3 moing window view into the array.
Requires:
---------
- a : an input 2d array. X and Y represent coordinates of the Z values
- cell_size : cell size, must be in the same units as X and Y
- kern : kernel to use
- degrees : True, returns degrees otherwise radians
- verb : True, to print results
- keep : False, to remove/squeeze extra dimensions
- filter :
np.array([[1, 2, 1], [2, 0, 2], [1, 2, 1]]) **current default
Notes:
------
::
dzdx: sum(col2 - col0)/8*cellsize
dzdy: sum(row2 - row0)/8*celsize
Assert the array is ndim=4 even if (1,z,y,x)
general dzdx + dzdy = dxyz
[[a, b, c], [[1, 0, 1], [[1, 2, 1] [[1, 2, 1]
[d, e, f] [2, 0, 2], + [0, 0, 0] = [2, 0, 2],
[g, h, i] [1, 0, 1]] [1, 2, 1]] [1, 2, 1]]
"""
frmt = """\n :----------------------------------------:
:{}\n :input array...\n {}\n :slope values...\n {!r:}
:----------------------------------------:
"""
# ---- stride the data and calculate slope for 3x3 sliding windows ----
np.set_printoptions(edgeitems=10, linewidth=100, precision=1)
a_s = stride(a, win=(3, 3), stepby=(1, 1))
if a_s.ndim < 4:
new_shape = (1,) * (4-len(a_s.shape)) + a_s.shape
a_s = a_s.reshape(new_shape)
#
kern = kernels(kern) # return the kernel if specified
# ---- default filter, apply the filter to the array ----
#
dz_dx, dz_dy = filter_a(a_s, a_filter=kern, cell_size=cell_size)
#
s = np.sqrt(dz_dx**2 + dz_dy**2)
if degrees:
s = np.rad2deg(np.arctan(s))
if not keep:
s = np.squeeze(s)
if verb:
p = " "
args = ["Results for slope_a... ",
indent(str(a), p), indent(str(s), p)]
print(dedent(frmt).format(*args))
return s
def slope_a(a, cell_size=1, degrees=True, verbose=False, keepdims=False):
"""Return slope in degrees for an input array using 3rd order
: finite difference method for a 3x3 moing window view into the array.
:
:Requires:
:--------
: a - an input 2d array. X and Y represent coordinates of the Z values
: cell_size - must be in the same units as X and Y
: degrees - True, returns degrees otherwise radians
: verbose - True, to print results
: keepdims - False, to remove/squeeze extra dimensions
: filter - np.array([[1, 2, 1], [2, 0, 2], [1, 2, 1]]) **current default
:
:Notes: dzdx: sum(col2 - col0)/8*cellsize
:----- dzdy: sum(row2 - row0)/8*celsize
: Assert the array is ndim=4 even if (1,z,y,x)
: general dzdx + dzdy = dxyz
: [[a, b, c], [[1, 0, 1], [[1, 2, 1] [[1, 2, 1]
: [d, e, f] [2, 0, 2], + [0, 0, 0] = [2, 0, 2],
: [g, h, i] [1, 0, 1]] [1, 2, 1]] [1, 2, 1]]
:
"""
frmt = """\n :----------------------------------------:
:{}\n :input array...\n {}\n :slope values...\n {!r:}
:----------------------------------------:
"""
# ---- stride the data and calculate slope for 3x3 sliding windows ----
np.set_printoptions(edgeitems=10, linewidth=100, precision=1)
a_s = stride(a, win=(3, 3), stepby=(1, 1))
if a_s.ndim < 4:
new_shape = (1, ) * (4 - len(a_s.shape)) + a_s.shape
a_s = a_s.reshape(new_shape)
f_dxyz = np.array([[1, 2, 1], [2, 0, 2], [1, 2, 1]], dtype="float64")
# ---- default filter, apply the filter to the array ----
#
dz_dx, dz_dy = filter_a(a_s, a_filter=f_dxyz, cell_size=cell_size)
#
s = np.sqrt(dz_dx**2 + dz_dy**2)
if degrees:
s = np.rad2deg(np.arctan(s))
if not keepdims:
s = np.squeeze(s)
if verbose:
from textwrap import indent, dedent # if not imported
p = " "
args = [
"Results for slope_a... ",
indent(str(a), p),
indent(str(s), p)
]
print(dedent(frmt).format(*args))
return s
def expand_zone(a, zone=None, win=2):
"""Expand a value (zone) in a 2D array, normally assumed to represent a
raster surface.
zone : number
The value/class to expand into the surrounding cells
win : list/tuple
select a (2, 2) or (3, 3) moving window
"""
msg = "\nYou need a zone that is within the range of values."
if zone is None:
print(msg)
return a, None
if (zone < a.min()) or (zone > a.max()):
print(msg)
return a, None
if win not in (2, 3):
win = 2
p = [1, 0][win == 2] # check for 2 or 3 in win
ap = np.pad(a, pad_width=(1, p), mode="constant", constant_values=(0, 0))
# n, m = ap.shape
if win == 2:
a_c = ap[1:, 1:] # for 2x2 even
elif win == 3:
a_c = ap[1:-1, 1:-1] # for 3x3 odd
a_s = stride(ap, win=(win, win), stepby=(win, win)) # stride the array
r, c = a_s.shape[:2]
out = []
x = a_s.shape[0]
y = a_s.shape[1]
for i in range(x):
for j in range(y):
if zone in a_s[i, j]:
out.append(1)
else:
out.append(0)
out1 = np.asarray(out).reshape(r, c)
out = np.repeat(np.repeat(out1, 2, axis=1), 2, axis=0)
dx, dy = np.array(out.shape) - np.array(a.shape)
if dx != 0:
out = out[:dx, :dy]
final = np.where(out == 1, zone, a_c)
return final
def expand_zone(a, zone=None, win=2):
"""Expand a value (zone) in a 2D array, normally assumed to represent a
raster surface.
zone : number
The value/class to expand into the surrounding cells
win : list/tuple
select a (2, 2) or (3, 3) moving window
"""
msg = "\nYou need a zone that is within the range of values."
if zone is None:
print(msg)
return a, None
if (zone < a.min()) or (zone > a.max()):
print(msg)
return a, None
if win not in (2, 3):
win = 2
p = [1, 0][win == 2] # check for 2 or 3 in win
ap = np.pad(a, pad_width=(1, p), mode="constant", constant_values=(0, 0))
# n, m = ap.shape
if win == 2:
a_c = ap[1:, 1:] # for 2x2 even
elif win == 3:
a_c = ap[1:-1, 1:-1] # for 3x3 odd
a_s = stride(ap, win=(win, win), stepby=(win, win)) # stride the array
r, c = a_s.shape[:2]
out = []
x = a_s.shape[0]
y = a_s.shape[1]
for i in range(x):
for j in range(y):
if zone in a_s[i, j]:
out.append(1)
else:
out.append(0)
out1 = np.asarray(out).reshape(r, c)
out = np.repeat(np.repeat(out1, 2, axis=1), 2, axis=0)
dx, dy = np.array(out.shape) - np.array(a.shape)
if dx != 0:
out = out[:dx, :dy]
final = np.where(out == 1, zone, a_c)
return final
def aspect_a(a, cell_size=1, flat=0.1, degrees=True, keepdims=False):
"""Return the aspect of a slope in degrees from North.
Requires:
--------
- a :
an input 2d array. X and Y represent coordinates of the Z values
- cell_size :
needed to proper flat calculation
- flat :
degree value, e.g. flat surface <= 0.05 deg
0.05 deg => 8.7e-04 rad 0.10 deg => 1.7e-02 rad
"""
if not isinstance(flat, (int, float)):
flat = 0.1
a_s = stride(a, win=(3, 3), stepby=(1, 1))
if a_s.ndim < 4:
new_shape = (1,) * (4-len(a_s.shape)) + a_s.shape
a_s = a_s.reshape(new_shape)
f_dxyz = np.array([[1, 2, 1], [2, 0, 2], [1, 2, 1]], dtype="float64")
a_s = a_s * f_dxyz
#
dz_dx, dz_dy = filter_a(a_s, a_filter=f_dxyz, cell_size=1)
#
asp = np.arctan2(dz_dy, -dz_dx) # relative to East
# get the slope
s = np.sqrt((dz_dx*cell_size)**2 + (dz_dy*cell_size)**2)
asp = np.rad2deg(asp)
asp = np.mod((450.0 - asp), 360.) # simplest way to get azimuth
asp = np.where(s <= flat, -1, asp)
if not keepdims:
asp = np.squeeze(asp)
if not degrees:
asp = np.deg2rad(asp)
return asp
def a_filter(a, mode=1, pad_output=True, ignore_nodata=True, nodata=None):
"""Various filters applied to an array.
Requires:
--------
stride : function
the `stride` function is used internally in this function
a : array
an array that will be strided using a 3x3 window
pad_output : boolean
True, produces a masked array padded so that the shape
is the same as the input
ignore_nodata : boolean
True, all values used, False, array contains nodata
nodata : None or number
None :
max int or float used
value :
use this value in integer or float form otherwise
mode :
a dictionary containing a choice from
::
1. `all_f` : all 1's
2. `no_cnt` : no center
3. `cross_f` : cross filter, corners
4. `plus_f` : up down, left right
5. `gradient` : `6 7 8 9 10` directional gradients
11. `lap_33` : laplacian
12. `line_h` : line detection, horizonal
13. `line_ld` : line detection, left diagonal
14. `line_rd` : line detection, right diagonal
15. `line_v` : line detection, vertical
16. `high` :
17. `sob_hor` : Sobel horizontal
18. `sob_vert` : Sobel vertical
19. `emboss` :
20. `sharp1` : sharpen 1
22. `sharp2` : sharpen 2
23. `sharp3` : sharpen 3 highpass 3x3
24. `lowpass` : lowpass filter
Notes:
-----
Only 3x3 filters covered here. The output array is padded with np.nan
and the array is returned as a masked array.
>>> a0 = pyramid(core=4, steps=5, incr=(1, 1))
>>> a0 = a0 * 2 # multiply by a number to increase slope
>>> a0 = (pyramid(core=4, steps=5, incr=(1, 1)) + 1) * 2 # is also good!
References:
----------
..
[1]
http://desktop.arcgis.com/en/arcmap/latest/tools/spatial-analyst-toolbox/how-filter-works.htm
[2]
http://desktop.arcgis.com/en/arcmap/latest/manage-data/raster-and-images/convolution-function.htm
[3]
https://github.com/scikit-image/scikit-image/tree/master/skimage/filters
"""
n = np.nan
all_f = [1, 1, 1, 1, 1, 1, 1, 1, 1]
no_cnt = [1, 1, 1, 1, n, 1, 1, 1, 1]
cross_f = [1, 0, 1, 0, 0, 0, 1, 0, 1]
plus_f = [0, 1, 0, 1, 0, 1, 0, 1, 0]
grad_e = [1, 0, -1, 2, 0, -2, 1, 0, -1]
grad_n = [-1, -2, -1, 0, 0, 0, 1, 2, 1]
grad_ne = [0, -1, -2, 1, 0, -1, 2, 1, 0]
grad_nw = [2, -1, 0, -1, 0, 1, 0, 1, 2]
grad_s = [1, 2, 1, 0, 0, 0, -1, -2, -1]
grad_w = [-1, 0, 1, -2, 0, 2, -1, 0, 1]
lap_33 = [0, -1, 0, -1, 4, -1, 0, -1, 0]
line_h = [-1, -1, -1, 2, 2, 2, -1, -1, -1]
line_ld = [2, -1, -1, -1, 2, -1, -1, -1, 2]
line_rd = [-1, -1, 2, -1, 2, -1, 2, -1, -1]
line_v = [-1, 0, -1, -1, 2, -1, -1, 2, -1]
high = [-0.7, -1.0, -0.7, -1.0, 6.8, -1.0, -0.7, -1.0, -0.7] # arc
sob_hor = [1, 2, 1, 0, 0, 0, -1, -2, -1] # sobel y /4.0 weights
sob_vert = [1, 0, -1, 2, 0, -2, 1, 0, -1] # sobel x /4
emboss = [-1, -1, 0, -1, 0, 1, 0, 1, 1]
sharp1 = [0., -0.25, 0., -0.25, 2.0, -0.25, 0., -0.25, 0.] # arc
sharp2 = [-0.25, -0.25, -0.25, -0.25, 3.0, -0.25, -0.25, -0.25, -0.25]
sharp3 = [-1, -1, -1, -1, 9, -1, -1, -1, -1] # arc
lowpass = [1, 2, 1, 2, 4, 2, 1, 2, 1] # arc
# ---- assemble the dictionary ----
d = {1: all_f, 2: no_cnt, 3: cross_f, 4: plus_f, 5: grad_e,
6: grad_n, 7: grad_ne, 8: grad_nw, 9: grad_s, 10: grad_w,
11: lap_33, 12: line_h, 13: line_ld, 14: line_rd, 15: line_v,
16: high, 17: sob_hor, 18: sob_vert, 19: emboss, 20: sharp1,
21: sharp2, 23: sharp3, 24: lowpass}
filter_ = np.array(d[mode]).reshape(3, 3)
# ---- stride the input array ----
a_strided = stride(a)
if ignore_nodata:
c = np.sum(a_strided * filter_, axis=(2, 3))
else:
c = np.nansum(a_strided * filter_, axis=(2, 3))
if pad_output:
pad_ = nodata
if nodata is None:
if c.dtype.name in ('int', 'int32', 'int64'):
pad_ = min([0, -1, a.min()-1])
else:
pad_ = min([0.0, -1.0, a.min()-1])
c = np.lib.pad(c, (1, 1), "constant", constant_values=(pad_, pad_))
m = np.where(c == pad_, 1, 0)
c = np.ma.array(c, mask=m, fill_value=None)
return c
def a_filter(a, mode=1, pad_output=True, ignore_nodata=True, nodata=None):
"""Various filters applied to an array.
Requires:
--------
stride : function
the `stride` function is used internally in this function
a : array
an array that will be strided using a 3x3 window
pad_output : boolean
True, produces a masked array padded so that the shape
is the same as the input
ignore_nodata : boolean
True, all values used, False, array contains nodata
nodata : None or number
None :
max int or float used
value :
use this value in integer or float form otherwise
mode :
a dictionary containing a choice from
::
1. `all_f` : all 1's
2. `no_cnt` : no center
3. `cross_f` : cross filter, corners
4. `plus_f` : up down, left right
5. `gradient` : `6 7 8 9 10` directional gradients
11. `lap_33` : laplacian
12. `line_h` : line detection, horizonal
13. `line_ld` : line detection, left diagonal
14. `line_rd` : line detection, right diagonal
15. `line_v` : line detection, vertical
16. `high` :
17. `sob_hor` : Sobel horizontal
18. `sob_vert` : Sobel vertical
19. `emboss` :
20. `sharp1` : sharpen 1
22. `sharp2` : sharpen 2
23. `sharp3` : sharpen 3 highpass 3x3
24. `lowpass` : lowpass filter
Notes:
-----
Only 3x3 filters covered here. The output array is padded with np.nan
and the array is returned as a masked array.
>>> a0 = pyramid(core=4, steps=5, incr=(1, 1))
>>> a0 = a0 * 2 # multiply by a number to increase slope
>>> a0 = (pyramid(core=4, steps=5, incr=(1, 1)) + 1) * 2 # is also good!
References:
----------
..
[1]
http://desktop.arcgis.com/en/arcmap/latest/tools/spatial-analyst-toolbox/how-filter-works.htm
[2]
http://desktop.arcgis.com/en/arcmap/latest/manage-data/raster-and-images/convolution-function.htm
[3]
https://github.com/scikit-image/scikit-image/tree/master/skimage/filters
"""
n = np.nan
all_f = [1, 1, 1, 1, 1, 1, 1, 1, 1]
no_cnt = [1, 1, 1, 1, n, 1, 1, 1, 1]
cross_f = [1, 0, 1, 0, 0, 0, 1, 0, 1]
plus_f = [0, 1, 0, 1, 0, 1, 0, 1, 0]
grad_e = [1, 0, -1, 2, 0, -2, 1, 0, -1]
grad_n = [-1, -2, -1, 0, 0, 0, 1, 2, 1]
grad_ne = [0, -1, -2, 1, 0, -1, 2, 1, 0]
grad_nw = [2, -1, 0, -1, 0, 1, 0, 1, 2]
grad_s = [1, 2, 1, 0, 0, 0, -1, -2, -1]
grad_w = [-1, 0, 1, -2, 0, 2, -1, 0, 1]
lap_33 = [0, -1, 0, -1, 4, -1, 0, -1, 0]
line_h = [-1, -1, -1, 2, 2, 2, -1, -1, -1]
line_ld = [2, -1, -1, -1, 2, -1, -1, -1, 2]
line_rd = [-1, -1, 2, -1, 2, -1, 2, -1, -1]
line_v = [-1, 0, -1, -1, 2, -1, -1, 2, -1]
high = [-0.7, -1.0, -0.7, -1.0, 6.8, -1.0, -0.7, -1.0, -0.7] # arc
sob_hor = [1, 2, 1, 0, 0, 0, -1, -2, -1] # sobel y /4.0 weights
sob_vert = [1, 0, -1, 2, 0, -2, 1, 0, -1] # sobel x /4
emboss = [-1, -1, 0, -1, 0, 1, 0, 1, 1]
sharp1 = [0., -0.25, 0., -0.25, 2.0, -0.25, 0., -0.25, 0.] # arc
sharp2 = [-0.25, -0.25, -0.25, -0.25, 3.0, -0.25, -0.25, -0.25, -0.25]
sharp3 = [-1, -1, -1, -1, 9, -1, -1, -1, -1] # arc
lowpass = [1, 2, 1, 2, 4, 2, 1, 2, 1] # arc
# ---- assemble the dictionary ----
d = {
1: all_f,
2: no_cnt,
3: cross_f,
4: plus_f,
5: grad_e,
6: grad_n,
7: grad_ne,
8: grad_nw,
9: grad_s,
10: grad_w,
11: lap_33,
12: line_h,
13: line_ld,
14: line_rd,
15: line_v,
16: high,
17: sob_hor,
18: sob_vert,
19: emboss,
20: sharp1,
21: sharp2,
23: sharp3,
24: lowpass
}
filter_ = np.array(d[mode]).reshape(3, 3)
# ---- stride the input array ----
a_strided = stride(a)
if ignore_nodata:
c = np.sum(a_strided * filter_, axis=(2, 3))
else:
c = np.nansum(a_strided * filter_, axis=(2, 3))
if pad_output:
pad_ = nodata
if nodata is None:
if c.dtype.name in ('int', 'int32', 'int64'):
pad_ = min([0, -1, a.min() - 1])
else:
pad_ = min([0.0, -1.0, a.min() - 1])
c = np.lib.pad(c, (1, 1), "constant", constant_values=(pad_, pad_))
m = np.where(c == pad_, 1, 0)
c = np.ma.array(c, mask=m, fill_value=None)
return c
