Description

A window function that returns a cluster number for each input geometry. Clustering partitions the geometries into sets in which each geometry is within the specified distance of at least one other geometry in the same cluster. Distances are Cartesian distances in the units of the SRID.

ST_ClusterWithinWin is equivalent to running ST_ClusterDBSCAN with minpoints => 0 .

Availability: 3.4.0

This method supports Circular Strings and Curves.

Examples

WITH testdata AS (
  SELECT id, geom::geometry FROM (
  VALUES  (1, 'LINESTRING (0 0, 1 1)'),
          (2, 'LINESTRING (5 5, 4 4)'),
          (3, 'LINESTRING (6 6, 7 7)'),
          (4, 'LINESTRING (0 0, -1 -1)'),
          (5, 'POLYGON ((0 0, 4 0, 4 4, 0 4, 0 0))')) AS t(id, geom)
)
SELECT id,
  ST_AsText(geom),
  ST_ClusterWithinWin(geom, 1.4) OVER () AS cluster
FROM testdata;


 id |           st_astext            | cluster
----+--------------------------------+---------
  1 | LINESTRING(0 0,1 1)            |       0
  2 | LINESTRING(5 5,4 4)            |       0
  3 | LINESTRING(6 6,7 7)            |       1
  4 | LINESTRING(0 0,-1 -1)          |       0
  5 | POLYGON((0 0,4 0,4 4,0 4,0 0)) |       0

        

See Also

ST_ClusterWithin , ST_ClusterDBSCAN , ST_ClusterIntersecting , ST_ClusterIntersectingWin ,