An Overview of Geospatial Data Structure, Algorithms, Mining, and Fusion

1
An Overview of Geospatial Data Structure,
Algorithms, Mining, and Fusion

• Presented by
• GDF Learning Group

2
Geospatial Data Structure

• Representation Modes
• Representing the Geometry of a Collection of
  Objects

3
Representation Modes of Geospatial Data Structure

• Tessellation
• Vector Mode
• Half-Plane Representation

4
Tessellation

• Fixed Tessellation (Regular)
• This model usually contains a grid or pattern.
• Ex. Raster data

• Variable Tessellation (Irregular)
• This model is less organized.
• Ex. An area partitioned into zones using polygons.

5
Vector Mode

• In this mode objects are constructed by points
  and edges. These are used to create a
  2-Dimensional plane.

• Structure Notation
• point
• polyline
• polygon
• region

6
Half-Plane Representation

• This type of modeling is used more in instances
  of 3D modeling.
• It combines the first two techniques and uses
  them to represent something more solid. Such as
  a building structure, mountains, valleys,etc.

7
Representing the Geometry of a Collection of
Objects

• Spaghetti Model
• Network Model
• Topological Model

8
Spaghetti Model

• Spaghetti model is a type of Vector model.
• Here, the geometry of any spatial object within
  is described independently from all others.
• The boundary of two adjacent regions is
  represented twice.
• The key to this model is simplicity.

9
Network Model

• Another type of Vector model.
• Designed to represent networks in network-based
  applications.
• In this type nodes are made the central points
  with arcs, lines, and polylines branching out.
• All polygons may not close.

10
Topological Model

• Another Vector model much like the network model.
• Only difference is that some nodes may not be
  connected by a line, arc, or polyline.
• Lines here are only stored once.

11
Algorithms

• In the GIS world

12
What is an algorithm?

• A procedure or formula for solving a problem.
  -www.cctvconsult.com/glossary.htm
• A finite set of step-by-step instructions for a
  problem-solving or computation procedure,
  especially one that can be implemented by a
  computer. -www.garlic.com/lynn/secgloss.htm

13
Types of algorithms in GIS

• Point in polygon
• Polyline intersections
• Polygon intersections
• Windowing
• Clipping

14
Point in Polygon

• Used to determine if a point lies in the area of
  a polygon
• Tests the edges of a polygon to see if the point
  lies on an edge of the polygon
• Tests the inside of the polygon to see if the
  point lies inside the polygon
• Refer to the text for the actual algorithm on
  page 178

15
(No Transcript)
16
Polyline intersections

• It is possible to detect intersections between
  polylines with algorithms
• Refer to the text for the actual algorithm on
  page 180
• It is also possible to compute new geometric
  objects as the result of an intersection between
  polylines
• Refer to the text for the actual algorithm on
  page 182

17
(No Transcript)
18
Polygon intersections

• Used to determine if two polygons overlap
• Checks if one polygon and another share an edge
• Also checks if one polygon is inside (or
  partially inside) another polygon
• Refer to the text for the actual algorithm on
  page 186

19
A
C
B
20
Windowing

• Tests whether a geometric object intersects a
  rectangle
• Also tests to see if the object is entirely
  contained by the rectangle
• Refer to the text for the actual algorithm on
  page 193

21
r
A
22
Clipping

• Computes the part of a geometric object that lies
  inside of a rectangle
• Refer to the text for the actual algorithm on
  page 196

23
(No Transcript)
24
Introduction to Geospatial Data Mining and
Knowledge Discovery

• Remote sensing technologies has greatly enhanced
  our capabilities to collect terabytes of
  geographic data
• This data is difficult to understand and needs to
  be transformed into useful and understandable
  information

25
Knowledge Discovery (KD) Technology

• Knowledge discovery technology empowers
  development of the next generation database
  management and info systems through its abilities
  to extract new, insightful info embedded within
  large heterogeneous databases and to formulate
  knowledge
• A KD process includes data warehousing, target
  data selection, cleaning, preprocessing,
  transformation and reduction, data mining, model
  selection (or combination), evaluation and
  interpretation, and consolidation and use of the
  extracted knowledge (Fayyad 1997, P5)

26
Knowledge Discovery

• Ultimately, KD aims to enable an information
  system to transform information to knowledge
  through hypothesis testing and theory formation

27
Data Mining (DM)

• Data mining is the non-trivial process of
  identifying valid, novel, potentially useful and
  ultimately understandable patterns in data
  (Fayyad et al. 1996)
• It aims to develop algorithms for extracting new
  patterns from the facts recorded in a database
• Data mining does not apply in cases where the
  outcome is already known

28
Data Mining

• The info data mining reveals should be useful and
  relevant
• Successful applications of DM are not common
• Establishing its relevance and explaining its
  cause are very difficult
• There are many analysis techniques that can be
  used to try to determine if info is useful

29
Data Mining

• Millions or even billions of hypotheses must be
  made
• It must be determined how false positives can be
  differentiated from truly significant findings

30
Visualization

• Discovered knowledge must be expressed in some
  way
• This data is best displayed visually so as to be
  better understood

31
Geospatial Data

• 3 characteristics of geospatial data create
  challenges to development of a robust data
  foundation

32
Characteristics of Geospatial Data

• Geospatial data repositories tend to be very
  large
• The second characteristic relates to phase
  characteristics of data collected cyclically
• Data discovery must accommodate collection cycles
  that may be unknown or that may shift from cycle
  to cycle in both time and space

33
Characteristics of Geospatial Data

• Third characteristic applies to a characteristic
  of the data foundation rather than of the data
• The internet has supported development of data
  clearinghouses, digital libraries, and online
  repositories wherein one does not access data,
  but pointers to data
• As digital data becomes more available on the
  internet, they become increasingly difficult to
  locate, retrieve, and analyze

34
Unique Properties of Geographic Data

• While KD applications involve highly dimensioned
  information spaces, geographic data is unique
  since up to four dimensions of the information
  space are interrelated and provide the
  measurement framework for the remaining dimensions

35
Unique Properties of Geographic Data

• Measured geographic attributes often exhibit the
  properties of spatial dependency and spatial
  heterogeneity
• Spatio-temporal objects and patterns are very
  complex

36
Unique Properties of Geographic Data

• The development of data mining and knowledge
  discovery tools must be supported by a solid
  geographic foundation that accommodates the
  unique characteristics and challenges presented
  by geospatial data

37
Geographic Knowledge Discovery Uses in Geographic
Research

• Map interpretation and info extraction
• Info extraction from remotely sensed imagery
• Mapping environmental features
• Extracting spatio-temporal patterns
• Spatial interaction, flow and movement in
  geographic space and human geographic systems

38
Critical Challenges in Geographic Knowledge
Discovery and Data Mining

• Developing and supporting geographic data
  warehouses
• Better spatio-temporal representations in
  geographic knowledge discovery
• Geographic KD using diverse data types
• User interfaces for geographic KD
• Proof of concepts and benchmarking
• Building discovered geographic knowledge into GIS
  and spatial analysis

39
Objectives

• Apply DM and KD techniques to the new generations
  of geospatial data models and identify analytical
  and visualizational needs for geospatial DM and
  KD
• Develop a taxonomy of geographic knowledge and
  categorize models for geographic information
  computing
• Enable a full implementation of geographic KD
  across distributed databases that allow the
  general public to inspect climate patterns and
  regional demographic dynamics, for example, on
  the internet

40
Geospatial Data Structure, Data Fusion

• Fusion the blending of 2 or more
  things.

41
Geospatial Data Fusion

• Fusion of gridded collections of measurement (ex.
  Image fusion)
• Fusion of remote sending image data and semantic
  data residing in a GIS

42
Image Fusion

• Data Level Fusion
• Feature Level Fusion
• Decision Level Fusion

43
Fusion of remote sending image data and semantic
data residing in a GIS

• Feature Level Fusion
• Decision Level Fusion
• Modeling of Processes (biogeochemical)

44
Data Level Fusion

• Spatial Domain Fusion
• Spectral Domain Fusion
• Scale-Space Fusion

45
Decision Level Fusion

• Classifier Fusion
• Classifier Selection

46
Conclusion

• Geospatial data integration requires a good
  understanding of data structure, algorithms, data
  mining, and knowledge discovery techniques.
• In our next discussion we will explore these
  concepts further.