Python Shapely – How to Add Points to a LineString in Shapely

pythonshapely

I have two Polygons which intersect along one edge as shown below.

I need to add the red vertices at each end of the intersection to the blue polygon. I thought I could do this with by calling union on the Polygon and the LineString produced by intersection of the two polygons but it appears not.

Is there a way of doing what I need to do?

Best Answer

You can use the properties of union (and unary_union) witch split all the lines at intersections (Planar graph)

# LinearRing of the polygons
ringgreen = LineString(list(polygreen.exterior.coords))
ringblue  = LineString(list(polyblue.exterior.coords))
# Union of the rings
union = ringgreen.union(ringblue)
# now polygonize the result
from shapely.ops import polygonize
result = [geom for geom in polygonize(union]

and

Related Solutions

[GIS] Adding a Property (attribute) to a geometry in Shapely/Fiona

Shapely geometries are python classes, so you can simply set a new property in the object.

For my use-cases this has worked, though it may be worth checking for unintended side-effects depending on your scenario.

from shapely import geometry

p = geometry.Point(0, 0)
p.uid = 'unique point'
print(p.uid)  # prints: 'unique point'

Shapely – Fast Intersection of Polygons with a Bounding Polygon

To speed up this kind of queries you have to use a spatial index. Use the STRtree class to build an rtree and exponentially speed up this operation. Here is the documentation with an example:

https://shapely.readthedocs.io/en/stable/manual.html#str-packed-r-tree

EDIT:

After seen the images, the STRtree will not be usefull at all. In this case, all the polygons intersect the bounding polygon so you need a different kind of optimization. I'll give you some ideas on how you could optimize this:

When you know a geometry must be tested against all (or almost all) the geometries in a set, you can 'prepare' this geometry to speed up testing operations. It don't supports 'overlaps' but you could use 'contains' to test the polygons that falls entirely within and 'intersects' for the boundary polygons.
Knowing that the polygons you are using are convex, instead of testing if the polygon is within the boundary polygon, you could test if the centroid is within, and if not test the intersection of the full polygon. If you don't know if they whould be convex, you could also use a representative point.
I'm thinking that, in some cases, the polygons that intersect the boundary could have the centroid within. You could avoid this by using a negative buffer of the boundary polygon.
You have a 'boundary polygon' but you could also have the boundary of this 'boundary polygon', wich is a linestring that only intersects the polygons that cross that line.
Also, you can extract the segments of the boundary polygon and use the STRtree to index it. Then iterate the polygons to see if they intersect any segment.

Maybe the most optimized way I know is to test if the centroid of the polygons are within a prepared negative buffer of the boundary polygon. If the centroid is not within, seems that you don't have polygons on the outside, but im not 100% sure so I'll give you two options:

If there are no polygons on the outside, then you know that if the centroid is not within it will intersect the boundary.
Else, you could test the intersection with the 'line boundary', or if it's not fast enougth, use the STRtree of the segments.

Best Answer

Related Solutions

[GIS] Adding a Property (attribute) to a geometry in Shapely/Fiona

Shapely – Fast Intersection of Polygons with a Bounding Polygon

Related Question