PostgreSQL/PostGIS：ST_distance 查询应该只返回“最近”的结果

Question

我想查询表 A 中每个点 (id) 到表 B 中每个多边形 (id) 的距离。对于距离计算，我使用 ST_Distance。但是，ST_Distance（显然）返回到每个多边形的距离。但我只需要每个点的“最接近”结果。到目前为止，我尝试了以下查询，它返回了正确的结果，但（当然）只针对一个点。

SELECT polygons.id, points.id, ST_Distance(points.geom, polygons.geom) 
FROM table_A AS points, table_B AS polygons 
ORDER BY st_distance ASC LIMIT 1

结果应该是这样的：

polygon_id | point_id | min(distance)
-------------------------------------
1234       |   876    |   54.32
...        |   ...    |    ...
-------------------------------------

你有什么提示吗？非常感谢。

更新 1

WITH CTE AS
       (SELECT polygons.id as poly_id, points.id as point_id, 
        ST_Distance(points.geom, polygons.geom) as thedistance , 
        row_number() OVER
              (PARTITION BY points.id ORDER BY ST_Distance(points.geom, polygons.geom)) 
    FROM 
        table_A AS points
    INNER JOIN table_B AS polygons
    ON ST_DWithin(points.geom, polygons.geom, 100)) SELECT * FROM CTE WHERE row_number = 1

运行上述查询（3h 24m）后返回空结果。但是，应该有结果。会不会是括号有问题？

更新 2

SRID 为 4326 (WGS84)，多边形为 OSM 构建多边形，点为同一城市中的任意点。

Answer 1

您可以使用window 函数和CTE。像这样的：

WITH CTE AS
   (select polygons.id as poly_id, points.id as point_id, 
    ST_Distance(points.geom, polygons.geom) as thedistance , 
    row_number() OVER
          (PARTITION BY points.id ORDER by ST_Distance(points.geom, polygons.geom) ) 
FROM 
 table_A AS points, table_B AS polygons )
 select poly_id, point_id, thedistance from CTE where row_number = 1

但是，如果您有很多积分，这可能会很慢。如果您大致知道点与多边形之间的距离，则可以通过在使用索引的连接中使用 st_dwithin 来加快速度。只需设置距离参数，以便捕捉每个点：

WITH CTE AS
   (select polygons.id as poly_id, points.id as point_id, 
    ST_Distance(points.geom, polygons.geom) as thedistance , 
    row_number() OVER
          (PARTITION BY points.id ORDER by ST_Distance(points.geom, polygons.geom) ) 
FROM 
    table_A AS points
INNER JOINT table_B AS polygons
ON st_Dwithin(points.geom, polygons.geom, 5000 )) -- assumes you have a metres projection, limit to 5KM
select * from CTE where row_number = 1

确保您的两个 GEOM 列上都有 GIST 索引