SDO_PC_PKG.CREATE_PC_UNIFIED

Format

CREATE_PC_UNIFIED(
  pc_type                  VARCHAR2,
  inp_table                VARCHAR2,
  base_table               VARCHAR2,
  data_table               VARCHAR2,
  pc_id                    NUMBER,
  pc_tol                   NUMBER,
  blk_size                 NUMBER,
  srid                     NUMBER       DEFAULT NULL,
  sfc_domain               SDO_GEOMETRY DEFAULT NULL,
  sfc_type                 VARCHAR2     DEFAULT '2D Hilbert as NUMBER', -- ('2D Hilbert as NUMBER', '2D Hilbert as RAW')
  pc_extent                SDO_GEOMETRY DEFAULT NULL,
  feature_table            VARCHAR2     DEFAULT NULL,
  feature_block_link_table VARCHAR2     DEFAULT NULL,
  feature_buffer           NUMBER       DEFAULT NULL,
  drop_non_interacting_pts NUMBER       DEFAULT 0,
  create_unblocked_table   NUMBER       DEFAULT 0,
  create_pyramid           NUMBER       DEFAULT 1,
  preserve_pyramid_leaves  NUMBER       DEFAULT 0
);

Description

Creates a point cloud of a configurable model (such as Hilbert R-tree blocked, R-tree blocked, Flat, or Hybrid).

Parameters

pc_type

Point cloud type or model.

Supported point cloud types are:

R-tree: Points are partitioned into geographic blocks, and this model is a variation of the R-tree structure.
Hilbert R-tree: Points are partitioned into geographic blocks, and this model is a variation of the Hilbert R-tree structure.
Feature-Blocked: Points are partitioned into geographic blocks, and this model is based on a given list of features (such as roads, postal codes, or building blocks).
Flat: Points are directly stored in a plain or flat database table. In this case, Exadata uses Exadata Smart Scan, while non-Exadata uses a B-tree on (VAL_D1, VAL_D2) or (VAL_D1, VAL_D2, VAL_D3). Also, on non-Exadata platforms, the flat model slows down queries beyond point cloud sizes of one billion points.
Hybrid or Hybrid Hilbert R-tree: Points are stored similarly to the flat model, except in an index-organized table using logical blocks similar to the Hilbert R-tree blocked model.

inp_table

Name of the table or view containing the input point cloud data.

This table or view must have the following columns:

VAL_D1 (NUMBER): Ordinate in dimension 1
VAL_D2 (NUMBER): Ordinate in dimension 2
...
VAL_Dn (NUMBER): Ordinate in dimension n, where n is the highest-numbered dimension. n must match the pc_tot_dimensions parameter value in the call to the SDO_PC_PKG.INIT function when the point cloud was initialized.

base_table

Name of the table in which the point cloud object will be stored.

The following shows an example DDL statement for creating this table:

CREATE TABLE pcs (id NUMBER, pc SDO_PC);

data_table

Name of the table in which the point cloud data will be stored.

The following shows an example DDL statement for creating this table for a blocked model:

CREATE TABLE blocks AS SELECT * FROM mdsys.sdo_pc_blk_table WHERE rownum < 0;

For a flat or hybrid model, the table will be automatically created in the database schema.

pc_id

Numerical point cloud ID.

pc_tol

Tolerance used in the point cloud.

blk_size

Size of a block in points.

This applies when R-tree blocked, Hilbert R-tree blocked, or Hybrid are used.

srid

SRID coordinate system used for the point cloud.

sfc_domain

Not currently used.

sfc_type

Not currently used.

pc_extent

Geographic extent of the point cloud.

feature_table

Name of the feature table used for feature blocking.

Points are blocked around the given features.

feature_block_link_table

Table created to provide the foreign key links between the feature table (feature_table) and the point cloud block table (data_table).

For example:

SQL> desc feature_block_links;
 Name                                      Null?    Type
 ----------------------------------------- -------- ---------
 FEATURE_ID                                         NUMBER
 OBJ_ID                                             NUMBER
 BLK_ID                                             NUMBER

feature_buffer

Geographic buffer around each feature, such as a highway.

drop_non_interacting_pts

Not currently used.

create_unblocked_table

Not currently used.

create_pyramid

Determines whether the point cloud can be organized in a pyramid, if blocked.

preserve_pyramid_leaves

Determines whether all points must be only stored in one of the pyramid blocks, or whether non-leaf points can also be stored in the leaves.

Usage Notes

The CREATE_PC_UNIFIED function supports all point cloud models, in contrast to CREATE_PC, which supports only the R-tree blocked model. The query equivalent to this function is SDO_PC_PKG.CLIP_PC_INTO_TABLE, which also supports all models. The function SDO_PC_PKG.CLIP_PC only works on blocked models.

Examples

The following example creates a Hilbert R-tree blocked point cloud with ID value 12345 and block size 10000.

call
  sdo_pc_pkg.create_pc_unified(
    pc_type    => 'Hilbert R-tree',
    inp_table  => 'INPTAB',
    base_table => 'PCS',
    data_table => 'BLOCKS',
    pc_id      => 12345,
    pc_tol     => 0.05,
    blk_size   => 10000,
    srid       => 27700,
    create_pyramid => 0
  );

The following example creates a feature-blocked point cloud. It is blocked based on the number of given features, rather than automatic clustering.

create table features (
  feature_id number,
  geom sdo_geometry);

call
  sdo_pc_pkg.create_pc_unified(
    pc_type                  => 'Feature-Blocked',
    inp_table                => 'INPTAB',
    base_table               => 'PCS',
    data_table               => 'BLOCKS',
    pc_id                    => 1,
    pc_tol                   => 0.05,
    blk_size                 => 10000,
    srid                     => null,
    feature_table            => 'FEATURES',
    feature_block_link_table => 'FEATURE_BLOCK_LINKS',
    feature_buffer           => 1,
    create_pyramid           => 0
  );