def sky_view_non_ditch_amplification(arr):
"""
Non ditch amplification from SkyViewFactor.
"""
arr = da.from_array(arr, chunks=(800, 800))
arr = d_gf(
arr, np.nanmedian,
footprint=create_circular_mask(25)).compute(scheduler='processes')
arr = da.from_array(_reclassify_sky_view_non_ditch_amp(arr),
chunks=(800, 800))
return d_gf(arr, np.nanmean, footprint=create_circular_mask(10))
def impoundment_amplification(arr, mask_radius=10):
"""
Impoundment ditch enhancement.
"""
norm_arr = da.from_array(_reclassify_impoundment(arr), chunks=(800, 800))
mask = create_circular_mask(mask_radius)
return d_gf(d_gf(d_gf(norm_arr, np.nanmean, footprint=mask),
np.nanmean,
footprint=mask),
np.nanmedian,
footprint=mask).compute(scheduler='processes')
def slope_non_ditch_amplification(arr):
"""
Non ditch amplification from Slope.
"""
new_arr = arr.copy()
arr = d_gf(
da.from_array(arr, chunks=(800, 800)),
np.nanmedian,
footprint=create_circular_mask(35)).compute(scheduler='processes')
new_arr = _slope_non_ditch_amplifcation_normalize(arr, new_arr)
return d_gf(da.from_array(new_arr, chunks=(800, 800)),
np.nanmean,
footprint=create_circular_mask(15))
def custom_remove_noise(arr, radius, threshold, selfThreshold):
"""
Removes noise from a probability prediction.
"""
max_arr = d_gf(
da.from_array(arr, chunks=(800, 800)),
np.nanmax,
footprint=general_functions.create_circular_mask(radius)).compute(
scheduler='processes')
return _custom_remove_noise(arr, max_arr, np.copy(arr), threshold,
selfThreshold)
def hpmf_filter(arr):
"""
HPMF ditch enhancement.
"""
normalized_arr = da.from_array(_reclassify_hpmf_filter(arr),
chunks=(800, 800))
mean = d_gf(
d_gf(d_gf(d_gf(normalized_arr,
np.amax,
footprint=create_circular_mask(1)),
np.amax,
footprint=create_circular_mask(1)),
np.median,
footprint=create_circular_mask(2)),
np.nanmean,
footprint=create_circular_mask(5)).compute(scheduler='processes')
reclassify = da.from_array(_reclassify_hpmf_filter_mean(mean),
chunks=(800, 800))
return d_gf(reclassify, np.nanmean, footprint=create_circular_mask(7))
def conic_proba_post_processing(arr, maskRadius, threshold):
"""
Attempts to fill the gaps of a continuous probability prediction array.
"""
masks = []
maxArr = d_gf(da.from_array(arr, chunks=(800, 800)),
np.nanmax,
footprint=general_functions.create_circular_mask(5))
for i in range(0, 8):
masks.append(create_conic_mask(maskRadius, i))
return _conic_proba_post_processing(np.array(arr), np.array(maxArr),
np.array(masks), threshold)
