删除 geopandas 中的相交线

Question

我有一个由 LineString 几何图形组成的大型地理数据框（可能有数百万行）。其中一些线可以相互相交，而另一些则不相交。我想以有效的方式删除所有相交的线。我有一个使用“apply”和一个 shapely 函数来查找交叉点的解决方案，但它在大型数据集上非常慢。任何帮助将不胜感激！

到目前为止，我使用一个小示例地理数据框的解决方案如下。这有效，但在放大时速度很慢。


from shapely.geometry import LineString
import geopandas as gpd
import numpy as np

def doesIntersect(line_A, line_B):
    return line_A.intersects(line_B)

def check_intersections(row, lines):
    # remove this row from the lines geoseries so we don't check against itself
    l = lines.drop(row.name)
    # return a boolean array to check if lines intersect
    intersects = [ doesIntersect(row['geometry'], x) for x in l ]

    # get the index of the first intersecting line
    idx = np.where(intersects)[0]

    # add the geometry of the intersecting line to the original geodataframe
    if not (len(idx) == 0):
        intersect_line = l.iloc[idx[0]]
        row['intersect_line'] = intersect_line
        print(intersect_line)
    else:
        row['intersect_line'] = 0
        print("no intersection")
    return row

# create an example gdf of intersecting lines
gdf = gpd.GeoDataFrame({'id': [1, 2, 3]}, geometry=[LineString([(1, 1), (4, 4)]),
                                                    LineString([(1, 4), (4, 1)]),
                                                    LineString([(6, 1), (6, 6)])])

# get the lines as a geoseries
lines = gdf['geometry']
# apply the function to the geodataframe
gdf = gdf.apply(check_intersections, axis=1, lines=lines)

# now drop intersecting lines
gdf = gdf[gdf['intersect_line'] == 0].drop(columns='intersect_line')
print(gdf)

Answer 1

利用内置 R 树空间索引的空间连接怎么样？

import geopandas as gpd
import pandas as pd


gdf = gpd.GeoDataFrame({'id': [1, 2, 3]}, geometry=[LineString([(1, 1), (4, 4)]),
                                                    LineString([(1, 4), (4, 1)]),
                                                    LineString([(6, 1), (6, 6)])])
# make a copy of the original gdf for spatial join
gdf_copy = gdf.copy(deep=True)
# spaital join the copy to the original gdf
gdf_sj = gpd.sjoin(gdf, gdf_copy)

gdf_sj_vc = gdf_sj['id_left'].value_counts()
# value counts for intersecting lines will be >=2 
int_mask = gdf_sj_vc.ge(2)
# add the mask to gdf_sj as intersects columnn
gdf_sj['intersects'] = gdf_sj['id_left'].map(int_mask)
# drop non-intersecting lines
gdf_no_intersect = gdf_sj.loc[gdf_sj['intersects'] == False]

print(gdf_no_intersect)

id_left geometry                                  index_right   id_right    intersects
2   3   LINESTRING (6.00000 1.00000, 6.00000 6.00000)          2    3       False

删除 geopandas 中的相交线

Removing intersecting lines in geopandas

python

shapely

geopandas