def dem_slope(self):
if (self._dem_data is None):
self.initialize_dem(1)
slope = dem.slope(self._dem_data)
file_name = self._create_file_name('slope', self._dem_data.num_aggre)
helper.create_tif(slope, file_name, self._orig_profile,
self._dem_data.num_aggre)
if (self.auto_plot):
helper.plot(file_name)
return file_name
def __init__(self,
file_path,
map_width_to_meters=1.0,
map_height_to_meters=1.0):
self._file_name = os.path.splitext(file_path)[0]
self._img = rasterio.open(file_path)
self._orig_profile = self._img.profile
self._dem_data = None
self.auto_plot = False
self.work_dtype = config.work_dtype
self.tif_dtype = config.tif_dtype
# Algorith is derived using assumption of square windows
# transform = self._img.profile['transform']
# self.pixel_width = math.sqrt(transform[0]**2 + transform[3]**2) * map_width_to_meters
# self.pixel_height = math.sqrt(transform[1]**2 + transform[4]**2) * map_height_to_meters
if (self.auto_plot):
helper.plot(file_path)
def dem_contour(self):
if (self._dem_data is None):
self.initialize_dem(1)
if (2**self._dem_data.num_aggre <= 2):
print(
'Profile curvature cannot be calculated on windows of size 2 or 1'
)
return
contour = dem.contour(self._dem_data)
file_name = self._create_file_name('contour', self._dem_data.num_aggre)
helper.create_tif(contour, file_name, self._orig_profile,
self._dem_data.num_aggre)
if (self.auto_plot):
helper.plot(file_name)
return file_name
def dem_profile(self):
if (self._dem_data is None):
self.initialize_dem(1)
if (2**self._dem_data.num_aggre <= 2):
print(
'Profile curvature cannot be calculated on windows of size 2 or 1'
)
return
profile = dem.profile(self._dem_data)
file_name = self._create_file_name('profile', self._dem_data.num_aggre)
# profile means two completely different things in the following line
helper.create_tif(profile, file_name, self._orig_profile,
self._dem_data.num_aggre)
if (self.auto_plot):
helper.plot(file_name)
return file_name
def multi_gauss(self, image_size=_image_size, experiment_id=0, noise=0.01):
if experiment_id == 0:
height_per_peak = 100
sigma_min = 0.1
sigma_spread = 0.2
random_set_size = 10
fn_trunc = 'multi_random'+'_id'+str(int(experiment_id))
elif experiment_id == 1:
height_per_peak = 10
sigma_min = 0.1
sigma_spread = 0.2
random_set_size = 20
fn_trunc = 'multi_random'+'_id'+str(int(experiment_id))
else:
experiment_id = 0
height_per_peak = 100
sigma_min = 0.1
sigma_spread = 0.2
random_set_size = 10
fn_trunc = 'multi_random'+'_id'+str(int(experiment_id))
if noise > 0:
fn_trunc = fn_trunc+'_n'+str(int(round(noise*1000)))
else:
fn_trunc = fn_trunc+'_nn'
random_set = np.empty((3,random_set_size))
rand.seed(1)
for i in range(random_set_size):
random_set[0,i] = rand.random()
random_set[1,i] = rand.random()
random_set[2,i] = rand.random()
print('Random point '+str(i)+' x = '+str(random_set[0,i])+' y = '+str(random_set[1,i])+' sigma factor = '+str(random_set[2,i]))
arr = np.zeros([image_size, image_size])
for y in range (image_size):
for x in range (image_size):
for peak in range(random_set_size):
mu = random_set[0,peak] * image_size
nu = random_set[1,peak] * image_size
sigma = (sigma_min + sigma_spread * random_set[2,peak]) * image_size
epsilon = noise * rand.random()
arr[x,y] += height_per_peak * (math.exp(-(((x - mu)**2 + (y - nu)**2)/(2 * sigma**2))) + epsilon)
# Original image
fn = self.path + fn_trunc + '.tif'
helper.create_tif(arr,fn)
if (self.auto_plot):
helper.plot(fn)
# Derivatives are shifted by half a pixel to make them comparable to the numerical results
fx_array = np.zeros([image_size, image_size])
fy_array = np.zeros([image_size, image_size])
fxx_array = np.zeros([image_size, image_size])
fyy_array = np.zeros([image_size, image_size])
fxy_array = np.zeros([image_size, image_size])
for y in range (image_size):
for x in range (image_size):
for peak in range(random_set_size):
mu = random_set[0,peak] * image_size
nu = random_set[1,peak] * image_size
sigma = (sigma_min + sigma_spread * random_set[2,peak]) * image_size
xs = x - 0.5
ys = y - 0.5
hill = math.exp(-(((xs - mu)**2 + (ys - nu)**2)/(2 * sigma**2)))
fx_array[x,y] += height_per_peak * ((mu - xs)/sigma**2 * hill)
fy_array[x,y] += height_per_peak * ((nu - ys)/sigma**2 * hill)
fxx_array[x,y] += height_per_peak * (((xs-mu)**2 - sigma**2)/sigma**4 * hill)
fyy_array[x,y] += height_per_peak * (((ys-nu)**2 - sigma**2)/sigma**4 * hill)
fxy_array[x,y] += height_per_peak * ((xs-mu)*(ys-nu)/sigma**4 * hill)
# Derivative
arr = np.arctan(np.sqrt(fx_array**2 + fy_array**2))*180/math.pi
fn = self.path + fn_trunc + '_slope.tif'
helper.create_tif(arr,fn)
if (self.auto_plot):
helper.plot(fn)
# Profile curvature
denom = fx_array**2 + fy_array**2
arr = -100*(fxx_array * fx_array**2 + 2 * fxy_array * fx_array * fy_array + fyy_array * fy_array**2) / denom
fn = self.path + fn_trunc + '_profile.tif'
helper.create_tif(arr,fn)
if (self.auto_plot):
helper.plot(fn)
# Tangential curvature
denom = fx_array**2 + fy_array**2
arr = -100*(fxx_array * fy_array**2 - 2 * fxy_array * fx_array * fy_array + fyy_array * fx_array**2) / denom
fn = self.path + fn_trunc + '_tangential.tif'
helper.create_tif(arr,fn)
if (self.auto_plot):
helper.plot(fn)
# Contour curvature
denom = (fx_array**2 + fy_array**2) * np.sqrt(fx_array**2+fy_array**2)
arr = -100*(fxx_array * fy_array**2 - 2 * fxy_array * fx_array * fy_array + fyy_array * fx_array**2) / denom
fn = self.path + fn_trunc + '_contour.tif'
helper.create_tif(arr,fn)
if (self.auto_plot):
helper.plot(fn)
# Proper Profile curvature
denom = (fx_array**2 + fy_array**2) * np.sqrt(1+(fx_array**2+fy_array**2))**3
arr = -100*(fxx_array * fx_array**2 + 2 * fxy_array * fx_array * fy_array + fyy_array * fy_array**2) / denom
fn = self.path + fn_trunc + '_proper_profile.tif'
helper.create_tif(arr,fn)
if (self.auto_plot):
helper.plot(fn)
# Proper Tangential curvature
denom = (fx_array**2 + fy_array**2) * np.sqrt(1+(fx_array**2+fy_array**2))
arr = -100*(fxx_array * fy_array**2 - 2 * fxy_array * fx_array * fy_array + fyy_array * fx_array**2) / denom
fn = self.path + fn_trunc + '_proper_tangential.tif'
helper.create_tif(arr,fn)
if (self.auto_plot):
helper.plot(fn)
return self.path+fn_trunc + '.tif'