Office of SA to CNS GeoIntelligence 2009. Introduction Data Mining vs Image Mining Image Mining - Issues and Challenges CBIR Image Mining Process Ontology.

Office of SA to CNSGeoIntelligence 2009

Introduction

Data Mining vs Image Mining

Image Mining - Issues and Challenges

CBIR

Image Mining Process

Ontology for Image Mining

Conclusion

GeoIntelligence 2009 Office of SA to CNS

An Overview of Image Mining

Image mining deals with the extraction of implicit knowledge, image data relationship, or other patterns not explicitly stored in the images.

Image mining is more than just an extension of data mining to image domain

Military Applications:Mobility Analysis

Traffficability Analysis

Potential Corridor of Landing

Image Mining

GeoIntelligence 2009 Office of SA to CNS

Office of SA to CNSGeoIntelligence 2009

The features may include: Color (in various channels), Texture (e.g. Directionality,

likeliness, contrast, roughness and coarseness), edge,

luminance, shape, spatial relations, temporal

information, statistical measures (e.g. moments – mean, variance, standard deviation, skewness etc).

Data mining searches: Valid patternsPreviously unknown patternsPotentially useful patternsUnderstandable patterns

Image mining extracts:Strategic informationRelationships and patternsLandscape aspects

Challenges (image mining)Relative valuesSpatial informationMultiple interpretationPatterns representation

GeoIntelligence 2009 Office of SA to CNS

Image information mining is an interdisciplinary endeavor Computer vision (image processing) Pattern recognition (classification & clustering) Databases (images & ancillary data) Information Retrieval (indexing and queries)

Challenges of mining information in remote sensing images Multi /hyper spectral (huge size, different formats) Variability of data sets (formats, types and structures) Time consuming preprocessing (correction and registration) Complex spatial / temporal associations Feature extraction & semantic definition (application specific) Ancillary data (climate variables, digital elevation model) Interpretation (a-priori and domain knowledge)

GeoIntelligence 2009 Office of SA to CNS

GeoIntelligence 2009 Office of SA to CNS

Content-based image retrieval (CBIR)

Modeling the contents of the image as a set of attributes

Using an integrated feature-extraction/object-recognition system

Image Mining Process

Image Databa

se

Pre-processing

Transformation & Feature Extraction

Mining

Interpretation & Evaluation

Knowledge

Office of SA to CNSGeoIntelligence 2009

Graph Mining Approach

Attribute Relational Graph (ARG)

Regional Adjacency Graph (RAG)

Ontological Approach

Image Mining Approaches

GeoIntelligence 2009 Office of SA to CNS

Method Ontology

Structural

Ontology

Physical Ontology

Semantic

Mediator

Application

Ontology

Task Ontology

Ontology for Image Mining

Incorporation of semantic information into the knowledge discovery process

Ontology describes a particular reality with a specific vocabulary, using a set of hypothesis related to the intentional meaning of the words in this vocabulary

Physical ontology

Structural ontology

Method Ontology

Office of SA to CNSGeoIntelligence 2009

Image Mining Systems

GeoMiner

A spatial data mining system developed by Han et al (1997)

ADaM

A NASA-developed Image Mining System

MSIM

A Multi-sensor Image Mining System developed by BAE Systems

GeoIntelligence 2009 Office of SA to CNS

GeoIntelligence 2009 Office of SA to CNS

Conclusion

Currently, most image processing techniques are designed to operate on a single image

Very few techniques for image data mining and information extraction in large image data sets

“Knowledge gap” in the process of deriving information from images and digital maps

Future research directions in remote sensing image mining include tracking individual trajectories of change

GeoIntelligence 2009 Office of SA to CNS

QUESTIONS

?

GeoIntelligence 2009 Office of SA to CNS

SA to CNS

Satellite Data

TERRAIN DATABASE MANAGEMENT SYSTEM (TDMS)

TDSSTerrain Decision Support System

Attribute Data, Spatial Data, & Knowledge Base

Spatial DB Server (GIS)

R DB Server (RDBMS)

Relational DB Miner

Spatial/ImageDB MinerTKDD

Terrain Knowledge Discovery from Data

Format Converter

TD&CS

VLDHVery Large

Database Handler

Military Applications

Interactive Mining I/F

Application Interface

DTDB

Terrain Analysis and Visualization

Training Set, Testing & Validation Data Set

MGD

TRMS

TPMS

Map Data

DATA INPUT

Expert Refinement

Discovered Rules/ Features (Natural & Manmade)

ADSApplication

Development System

Knowledge Acquisition I/F

Domain Expert

Knowledge Base

Knowledge compiler

GIS Mapping I/F

Inference Mechanism

PCLNMOCCM

Field Data

Scale Converter

Projection Converter

DTSData Transformation System

GIS Mapping I/F

DTRL

CCM-Cross Country Mobility

NMO-Natural & Manmade Obstacles

PCL-Potential Corridor of Landing

TD&CS-Troops Deployment & Camping Sites

LOS-Line of Sight

LOS