def centroid_corner(geom):
'''all these characters working with corners could be merged and cleaned
'''
from shapely.geometry import Point
if geom is None:
return (None, None)
distances = [] # set empty list of distances
centroid = geom.centroid # define centroid
points = list(geom.exterior.coords) # get points of a shape
stop = len(points) - 1 # define where to stop
for i in np.arange(
len(points)
): # for every point, calculate angle and add 1 if True angle
if i == 0:
continue
elif i == stop:
a = np.asarray(points[i - 1])
b = np.asarray(points[i])
c = np.asarray(points[1])
p = Point(points[i])
if _true_angle(a, b, c) is True:
distance = centroid.distance(
p
) # calculate distance point - centroid
distances.append(distance) # add distance to the list
else:
continue
else:
a = np.asarray(points[i - 1])
b = np.asarray(points[i])
c = np.asarray(points[i + 1])
p = Point(points[i])
if _true_angle(a, b, c) is True:
distance = centroid.distance(p)
distances.append(distance)
else:
continue
if not distances: # circular buildings
from momepy.dimension import _longest_axis
if geom.has_z:
coords = [
(coo[0], coo[1]) for coo in geom.convex_hull.exterior.coords
]
else:
coords = geom.convex_hull.exterior.coords
return (_longest_axis(coords) / 2, 0)
return (np.mean(distances), np.std(distances))
def __init__(self, gdf):
self.gdf = gdf
# define empty list for results
results_list = []
results_list_sd = []
# calculate angle between points, return true or false if real corner
def true_angle(a, b, c):
ba = a - b
bc = c - b
cosine_angle = np.dot(
ba, bc) / (np.linalg.norm(ba) * np.linalg.norm(bc))
angle = np.arccos(cosine_angle)
if np.degrees(angle) <= 170:
return True
if np.degrees(angle) >= 190:
return True
return False
# iterating over rows one by one
for index, row in tqdm(gdf.iterrows(), total=gdf.shape[0]):
distances = [] # set empty list of distances
centroid = row["geometry"].centroid # define centroid
points = list(
row["geometry"].exterior.coords) # get points of a shape
stop = len(points) - 1 # define where to stop
for i in np.arange(
len(points)
): # for every point, calculate angle and add 1 if True angle
if i == 0:
continue
elif i == stop:
a = np.asarray(points[i - 1])
b = np.asarray(points[i])
c = np.asarray(points[1])
p = Point(points[i])
if true_angle(a, b, c) is True:
distance = centroid.distance(
p) # calculate distance point - centroid
distances.append(distance) # add distance to the list
else:
continue
else:
a = np.asarray(points[i - 1])
b = np.asarray(points[i])
c = np.asarray(points[i + 1])
p = Point(points[i])
if true_angle(a, b, c) is True:
distance = centroid.distance(p)
distances.append(distance)
else:
continue
if not distances: # circular buildings
from momepy.dimension import _longest_axis
results_list.append(
_longest_axis(row["geometry"].convex_hull.exterior.coords)
/ 2)
results_list_sd.append(0)
else:
results_list.append(np.mean(distances)) # calculate mean
results_list_sd.append(np.std(distances)) # calculate st.dev
self.mean = pd.Series(results_list, index=gdf.index)
self.std = pd.Series(results_list_sd, index=gdf.index)
def centroid_corners(objects):
"""
Calculates mean distance centroid - corners and st. deviation.
.. math::
\\overline{x}=\\frac{1}{n}\\left(\\sum_{i=1}^{n} dist_{i}\\right);\\space \\mathrm{SD}=\\sqrt{\\frac{\\sum|x-\\overline{x}|^{2}}{n}}
Parameters
----------
objects : GeoDataFrame
GeoDataFrame containing objects
Returns
-------
means, st_devs : tuple
means : Series
Series containing mean distance values.
st_devs : Series
Series containing standard deviation values.
References
----------
Schirmer PM and Axhausen KW (2015) A multiscale classification of urban morphology.
Journal of Transport and Land Use 9(1): 101–130.
+ Cimburova (ADD)
Examples
--------
>>> means, deviations = momepy.centroid_corners(buildings_df)
Calculating distance centroid - corner...
100%|██████████| 144/144 [00:00<00:00, 846.58it/s]
Distances centroid - corner calculated.
>>> buildings_df['ccd_means'] = means
>>> buildings_df['ccd_stdev'] = deviations
>>> buildings_df['ccd_means'][0]
15.961531913184833
>>> buildings_df['ccd_stdev'][0]
3.0810634305400177
"""
# define empty list for results
results_list = []
results_list_sd = []
print('Calculating distance centroid - corner...')
# calculate angle between points, return true or false if real corner
def true_angle(a, b, c):
ba = a - b
bc = c - b
cosine_angle = np.dot(ba,
bc) / (np.linalg.norm(ba) * np.linalg.norm(bc))
angle = np.arccos(cosine_angle)
if np.degrees(angle) <= 170:
return True
elif np.degrees(angle) >= 190:
return True
return False
# iterating over rows one by one
for index, row in tqdm(objects.iterrows(), total=objects.shape[0]):
distances = [] # set empty list of distances
centroid = row['geometry'].centroid # define centroid
points = list(row['geometry'].exterior.coords) # get points of a shape
stop = len(points) - 1 # define where to stop
for i in np.arange(
len(points)
): # for every point, calculate angle and add 1 if True angle
if i == 0:
continue
elif i == stop:
a = np.asarray(points[i - 1])
b = np.asarray(points[i])
c = np.asarray(points[1])
p = Point(points[i])
if true_angle(a, b, c) is True:
distance = centroid.distance(
p) # calculate distance point - centroid
distances.append(distance) # add distance to the list
else:
continue
else:
a = np.asarray(points[i - 1])
b = np.asarray(points[i])
c = np.asarray(points[i + 1])
p = Point(points[i])
if true_angle(a, b, c) is True:
distance = centroid.distance(p)
distances.append(distance)
else:
continue
if not distances:
from momepy.dimension import _longest_axis
results_list.append(
_longest_axis(row['geometry'].convex_hull.exterior.coords) / 2)
results_list_sd.append(0)
else:
results_list.append(np.mean(distances)) # calculate mean
results_list_sd.append(np.std(distances)) # calculate st.dev
means = pd.Series(results_list)
st_devs = pd.Series(results_list_sd)
print('Distances centroid - corner calculated.')
return means, st_devs