Spatial indexing and querying

Large Databases
Week 6

Māra Pudāne
[email protected]

10.10.2023.

1
 Geometry validation

Rīgas Tehniskā universitāte 2

Validating spatial data
▪ Enables checking whether your input is correct
▪ Done using
SDO_GEOM.VALIDATE_GEOMETRY_WITH_CONTEXT and
SDO_GEOM.VALIDATE_LAYER_WITH_CONTEXT subprograms

3
General procedure
▪ Loading data
▪ Running
SDO_GEOM.VALIDATE_GEOMETRY_WITH_CONTEXT or
SDO_GEOM.VALIDATE_LAYER_WITH_CONTEXT
▪ Use SDO_MIGRATE.TO_CURRENT to rectify geometry
according to current version
▪ Use SDO_UTIL.RECTIFY_GEOMETRY to fix geometries
▪ However, most work must be done manually

4
Checks type consistency and geometry
consisency
▪ For type consistency, the function checks
– It the SDO_GTYPE is valid
– If the SDO_ETYPE values are consistent with the SDO_GTYPE value
– If the SDO_ELEM_INFO_ARRAY has valid triplet values
▪ For geometry consistency, the function checks
– Polygons have at least four points, which includes the point that closes the poly
– Polygons are not self-crossing
– No two vertices on a line or polygon are the same
– Polygons are oriented correctly
– Points on an arc are not colinear
– Geometries are within the specified bounds of the applicable DIMINFO column value
– etc

Rīgas Tehniskā universitāte 5

SDO_GEOM.VALIDATE_GEOMETRY_WI
TH_CONTEXT
▪ If the geometry is valid, this function returns TRUE
▪ If the geometry is not valid, this function returns the following:
– An Oracle error message number based on the specific reason the geometry is invalid, or FALSE if
the geometry fails for some other reason
– The context of the error (the coordinate, edge, or ring that causes the geometry to be invalid)

Rīgas Tehniskā universitāte 6

Validation – a quick example
▪ SELECT food_deliveries.my_title,
SDO_GEOM.VALIDATE_GEOMETRY_WITH_CONTEXT
(food_deliveries.shape, 0.5)
FROM food_deliveries;

Rīgas Tehniskā universitāte 7

 Example

Rīgas Tehniskā universitāte 8

 Spatial indexing

Rīgas Tehniskā universitāte 9

Spatial indexing is necessary
▪ Spatial indices are used by spatial databases to optimize
spatial queries
▪ Conventional index types do not efficiently handle spatial queries
such as how far two points differ, or whether points fall within a
spatial area of interest
– To find objects within an indexed data space that interact with a
given point or area of interest (window query)
– To find pairs of objects from within two indexed data spaces that
interact spatially with each other (spatial join)

10
Spatial data indexes

Methods for indexing

spatial data

Space-oriented
indexing Data-oriented
indexing
Grid
R-tree
Quadtree
Etc.
etc.

Rīgas Tehniskā universitāte 11

Decomposing the space/plane
▪ The index-creation process decomposes the space into a four-level grid
hierarchy
▪ This grid (or, tiles) are used to approximate geometries

▪ Tesselation rules:
– If an object completely covers a cell, that cell is said to be covered by the object.
– The extent of tessellation of each object depends primarily on the cells-per-object limit of
the spatial index. This limit defines the maximum number of cells that tessellation can
count per object.
– The deepest-cell rule exploits the fact that every lower-level cell belongs to the cell above
it. Therefore, only the deepest-level cells need to be recorded in the index,
minimizing the information that the index needs to store

12
The most common index
▪ B-trees, short for balanced
trees, are the most common
type of database index
▪ B-tree index is an ordered list
of values divided into ranges

Rīgas Tehniskā universitāte 13

Grid
▪ A grid is a regular tesselation or 2-D surface
that divides it into a series of contiguous cells,
which can then be assigned unique identifiers
and used for spatial indexing purposes.

▪ "square" or "rectangular" cells

▪ triangular grids or meshes,
▪ hexagonal grids
▪ etc.

Rīgas Tehniskā universitāte 14

Quadtree
▪ A quadtree is a tree data
structure in which each
internal node has exactly four
children
▪ Quadtrees are most often
used to partition a two-
dimensional space by
recursively subdividing it into
four quadrants or regions.

Rīgas Tehniskā universitāte 15

Octree
▪ Similar to quadtree but
used for 3D space
▪ Often used in 3D graphics
and 3D game engines

Rīgas Tehniskā universitāte 16

R-tree
▪ R-tree. Typically the preferred
method for indexing spatial
data. Objects (shapes, lines
and points) are grouped using
the MBR)

Rīgas Tehniskā universitāte 17

Creating index in Oracle
CREATE INDEX IND_BUILD ON BUILDINGS(BUILDING)
INDEXTYPE IS MDSYS.SPATIAL_INDEX;

CREATE INDEX cola_spatial_idx

ON cola_markets(shape)
INDEXTYPE IS MDSYS.SPATIAL_INDEX
PARAMETERS ('layer_gtype=POLYGON');

Rīgas Tehniskā universitāte 18

 Querying spatial data

Rīgas Tehniskā universitāte 19

The two-tier query model

20
Several perspectives
▪ One can query:
– Primary filter
– Secondary filter...

▪ ...and retrieve
– Topological information
– Information on geometries (such as area)

▪ Additionally:
– Querying the non-geometric data

21
Spatial query window

22
 Data retrieval from the
primary filter

Rīgas Tehniskā universitāte 23

SDO_FILTER
▪ Retrieves geometries, that are somehow related to area
▪ The primary filter checks to see if the MBRs of the candidate objects interact, not whether the
objects themselves interact

SDO_FILTER
(geometry1 SDO_GEOMETRY,
geometry2 SDO_GEOMETRY,
param VARCHAR2)

Rīgas Tehniskā universitāte 24

Example (1)
Retrieve all geometries related to (4,6), (8,8) dynamic query window

SELECT C.NUM, C.TITLE

FROM Layer_1 C
WHERE SDO_FILTER(C.Geometr, SDO_GEOMETRY(2003, NULL,
NULL, SDO_ELEM_INFO_ARRAY(1,1003,3),
SDO_ORDINATE_ARRAY(4,6, 8,8)) ) = 'TRUE’;

C.NUM C.TITLE
--------------------------------------
5 Geom 5
7 Geom 7

Rīgas Tehniskā universitāte 25

Example (2)
Retrieve all geometries related to windows stored in a table WINDOWS

SELECT A.Feature_ID FROM TARGET A, WINDOWS B

WHERE B.ID = 'WINS_1' AND
sdo_filter(A.shape, B.shape) = 'TRUE';

Rīgas Tehniskā universitāte 26

Example (3)
▪ Get geometries from Layer_1, that are related to geometry G1 from Layer_2

SELECT A.Title
FROM Layer_1 A, Layer_2 B
WHERE B.Title = ‘G1’ AND SDO_FILTER(A.Geometry,
B.Geometry) = 'TRUE';

Rīgas Tehniskā universitāte 27

Secondary filter querying

28
Secondary filter methods
▪ SDO_JOIN: Performs a spatial join based on one or more topological relationships.
▪ SDO_NN: Determines the nearest neighbour geometries to a geometry.
▪ SDO_NN_DISTANCE: Returns the distance of an object returned by the SDO_NN
operator.
▪ SDO_RELATE: Determines whether or not two geometries interact in a specified way.
▪ SDO_WITHIN_DISTANCE: Determines if two geometries are within a specified distance
from one another.

+
More specific filters, such as SDO_EQUAL, etc.

Rīgas Tehniskā universitāte 29

SDO_RELATE
▪ Uses 9-intersection matrix:
– Interior
– Boundary
– Exterior

▪ Retrieves binary topological relationships

by manipulating this matrix

30
Binary topological relationships

Rīgas Tehniskā universitāte 31

Example
SELECT A.Feature_ID FROM TARGET A
WHERE sdo_relate(A.shape, SDO_geometry(2003,NULL,NULL,
SDO_elem_info_array(1,1003,3),
SDO_ordinate_array(x1,y1, x2,y2)), 'mask=anyinteract') = 'TRUE’;

▪ How to implement OR:

'mask=inside+touch'

Rīgas Tehniskā universitāte 32

More specific filters
▪ Provided for conveinience
▪ Perform SDO_RELATE on a specific mask type

SDO_ANYINTERACT(geometry1, geometry2);

Rīgas Tehniskā universitāte 33

SDO_WITHIN_DISTANCE
▪ Used to determine the set of objects in a table that are within n distance units from a
reference object
▪ Only 2D

SDO_WITHIN_DISTANCE
(geometry1 SDO_GEOMETRY,
aGeom SDO_GEOMETRY,
params VARCHAR2);

Rīgas Tehniskā universitāte 34

Example
SELECT A.NUM FROM Layer_1 A
WHERE
SDO_WITHIN_DISTANCE(A.Geometry, sdo_geometry(2003,NULL,NULL,
sdo_elem_info_array(1,1003,3), sdo_ordinate_array(x1,y1,x2,y2)),
'distance = 10') = 'TRUE';

Rīgas Tehniskā universitāte 35

Example with two layers
SELECT A.NUM FROM Layer_1 A, Layer_2 B
WHERE B.NUM = 1 AND
SDO_WITHIN_DISTANCE(A.Geometry_1, B.Geometry_2, 'distance = 10') = 'TRUE';

Rīgas Tehniskā universitāte 36

SDO_NN
▪ Used to identify the nearest neighbours for a geometry
▪ Can be used only if a spatial index has been created on two dimensions of data

SDO_NN
(geometry1 SDO_GEOMETRY,
geometry2 SDO_GEOMETRY,
param VARCHAR2 [, number NUMBER]);

Rīgas Tehniskā universitāte 37

Example
SELECT r.name FROM restaurants r WHERE
SDO_NN(r.geometry, :my_hotel,
'sdo_batch_size=10 distance=2 unit=mile') = 'TRUE'
AND r.cuisine = 'Italian' AND ROWNUM <=2;

Rīgas Tehniskā universitāte 38

Spatial join SDO_JOIN
▪ A spatial join takes place when you compare all geometries of one layer to all geometries of
another layer.
▪ SDO_JOIN is technically not an operator, but a table function. This function is
recommended when you need to perform full table joins.

▪ Returns an object of SDO_ROWIDSET, which consists of a table of objects of

SDO_ROWIDPAIR.

Rīgas Tehniskā universitāte 39

Primary filter join
SELECT /*+ ordered */ a.gid, b.gid
FROM TABLE(SDO_JOIN('PARKS', 'SHAPE',
'HIGHWAYS', 'SHAPE',
'mask=FILTER')) c,
parks a,
highways b
WHERE c.rowid1 = a.rowid AND c.rowid2 = b.rowid;

Rīgas Tehniskā universitāte 40

Secondary filter join
SELECT /*+ ordered */ a.gid, b.gid
FROM TABLE(SDO_JOIN('PARKS', 'SHAPE',
'HIGHWAYS', 'SHAPE',
'mask=ANYINTERACT')) c,
parks a,
highways b
WHERE c.rowid1 = a.rowid AND c.rowid2 = b.rowid;

Rīgas Tehniskā universitāte 41

 Spatial aggregate functions

Rīgas Tehniskā universitāte 42

Various spatial aggregate functions
▪ SDO_AGGR_CENTROID: Returns a geometry object that is the centroid ("center of gravity")
of the specified geometry objects.
▪ SDO_AGGR_CONCAT_LINES: Returns a geometry that concatenates the specified line or
multiline geometries.
▪ SDO_AGGR_CONVEXHULL: Returns a geometry object that is the convex hull of the
specified geometry objects.
▪ SDO_AGGR_LRS_CONCAT: Returns an LRS geometry object that concatenates specified
LRS geometry objects.
▪ SDO_AGGR_MBR: Returns the minimum bounding rectangle of the specified geometries.
▪ SDO_AGGR_SET_UNION: Takes a VARRAY of SDO_GEOMETRY objects as input, and
returns the aggregate union of all geometry objects in the array.
▪ SDO_AGGR_UNION: Returns a geometry object that is the topological union (OR operation)
of the specified geometry objects.

43
SDO_GEOM functions

44
Operations
▪ Relationship (True/False) between two objects: RELATE,
WITHIN_DISTANCE
▪ Validation: VALIDATE_GEOMETRY_WITH_CONTEXT,
VALIDATE_LAYER_WITH_CONTEXT
▪ Single-object operations: SDO_ALPHA_SHAPE,
SDO_ARC_DENSIFY, SDO_AREA, SDO_BUFFER,
SDO_CENTROID, SDO_CONVEXHULL, SDO_CONCAVEHULL,
SDO_CONCAVEHULL_BOUNDARY, SDO_LENGTH,
SDO_MAX_MBR_ORDINATE, SDO_MIN_MBR_ORDINATE,
SDO_MBR, SDO_POINTONSURFACE, SDO_TRIANGULATE,
SDO_VOLUME
▪ Two-object operations: SDO_CLOSEST_POINTS, SDO_DISTANCE,
SDO_DIFFERENCE, SDO_INTERSECTION, SDO_UNION,
SDO_XOR
45
SDO_GEOM.SDO_AREA

SELECT c.name, SDO_GEOM.SDO_AREA(c.shape, 0.005) FROM

cola_markets c
WHERE c.name = 'cola_a';

Rīgas Tehniskā universitāte 46

SDO_GEOM.SDO_CLOSEST_POINTS
SDO_GEOM.SDO_CLOSEST_POINTS(
geom1 IN SDO_GEOMETRY,
geom2 IN SDO_GEOMETRY,
tolerance IN NUMBER,
unit IN VARCHAR2,
dist OUT NUMBER,
geoma OUT SDO_GEOMETRY,
geomb OUT SDO_GEOMETRY);

Rīgas Tehniskā universitāte 47

SDO_GEOM.SDO_CLOSEST_POINTS (1)
DECLARE
cola_c_geom SDO_GEOMETRY;
cola_d_geom SDO_GEOMETRY;
dist NUMBER;
geoma SDO_GEOMETRY;
geomb SDO_GEOMETRY;
BEGIN
SELECT c.shape into cola_c_geom FROM cola_markets c
WHERE c.name = 'cola_c';
SELECT c.shape into cola_d_geom FROM cola_markets c
WHERE c.name = 'cola_d';

Rīgas Tehniskā universitāte 48

SDO_GEOM.SDO_CLOSEST_POINTS (2)
SDO_GEOM.SDO_CLOSEST_POINTS(cola_c_geom,
cola_d_geom, 0.005, NULL,
dist, geoma, geomb);
INSERT INTO cola_markets VALUES(9901, 'geoma', geoma);
INSERT INTO cola_markets VALUES(9902, 'geomb', geomb);
DBMS_OUTPUT.PUT_LINE('dist output parameter value = ' ||
dist);
END;

Rīgas Tehniskā universitāte 49

SDO_GEOM.SDO_LENGTH

SELECT c.name, SDO_GEOM.SDO_LENGTH(c.shape, m.diminfo)

FROM cola_markets c, user_sdo_geom_metadata m
WHERE m.table_name = 'COLA_MARKETS' AND m.column_name = 'SHAPE';

Rīgas Tehniskā universitāte 50

Visualisation

51
Geometry validation in a tool

52
Simple visualization

Rīgas Tehniskā universitāte 53

View/map view

Rīgas Tehniskā universitāte 54

Visualisation with GeoRaptor*

*GeoRaptor might not work with newer versions of SQLDeveloper

Rīgas Tehniskā universitāte 55

After installing

Rīgas Tehniskā universitāte 56

Visualization with programming
language

Rīgas Tehniskā universitāte 57

Practical work
▪ Select the topic for individual work
– Example: Spatial data model in Oracle
▪ Write a theoretical background on the topic with
references (3-4 pages, master thesis formatting)!
▪ Do a practical work – depends on the topic!
– Example:
– Define 10 shapes and three layers!
– Write 8 spatial queries!
Rīgas Tehniskā universitāte 58
Class work

▪ Split into groups of 2-3 (make sure there is at least one

with computer), you can also work alone!

▪ Write a primary and secondary filter query based on the

valid geometries in the script!

Rīgas Tehniskā universitāte 59

Q&A

60