Skip to main content

Forest image processing supported by expert system

Objective

To develop an expert system to be used as a decision tool towards
protecting forest land.


The proposed expert system will be
based on remotely sensed Multi-date Satellite Imagery,
Geographical Information Systems, Global Positioning Signals,
Mathematical Modeling and Advanced Visualization Techniques.

The system will consist of the following main
parts: (i) an electronic database within a Geographical
Information System to provide information on forest areas, (ii) a
mathematical model to process the database information and
classify forested areas according to their potentiality for fire
eruption, (iii) a visualization system to provide an on-line
pixmap representation of the potential of fire eruption on a
digital map of the area of interest. To construct the expert
system, representative forest-fire affected areas from two
Mediterranean countries, will be studied.

The electronic database will be incorporated, structured and
layered in a Geographical Information System (EASI/PACE) and will
include: multi-date satellite data, digitized geophysical data,
land-use data and socio-economic parameters pertaining to factors
that affect intentional fire eruption.

Image processing and photo-interpretation techniques will be
applied on geophysical maps and multi-date satellite imagery
(LANDSAT TM, LANDSAT TM+ERS-1 SAR, LANDSAT TM+SPOT PAN) to
determine geophysical parameters such as: (I) biomass, (II) soil
moisture, and land use data such as: (I) illegal residential
development across forest roads and at the outskirts of the
existing forest areas, and (II) intensity of illegal housing
construction. Change detection techniques such as: Post
Classification Comparison, Principal Component Analysis (PCA) and
Image Rationing/Differencing will be applied on multi-date
satellite data of fire-affected areas, to determine land-use
change data.

Critical parameters will be determined, by cross-correlating
geophysical, land-use change, and socio-economic data pertaining
to areas that have been affected by fire eruption. These
parameters will serve as input to the mathematical model. A
deductive neural network classification algorithm will be
developed to classify forested areas according to their
potentiality for fire eruption.

The mathematical model will be integrated with the GIS system
(EASI/PACE) to provide automatic access to the input parameters
residing at the GIS database. Also, a visualization software
package (AVS) will be used to integrate the mathematical model
with advanced visualization techniques.

Funding Scheme

CSC - Cost-sharing contracts

Coordinator

INTEGRATED INFORMATION SYSTEMS SA
Address
72-74,Salaminos Street
17675 Kallithea
Greece

Participants (4)

Advanced Technology Information Systems
Greece
Address
5,Agiou Nikolaou
15772 Athens
CENTRO NACIONAL DE INFORMACAO GEOGRAFICA
Portugal
Address
82,Tagus Park, Nucleo Central Sala 301
2780-920 Oeiras
Centro di Ricerche e Servizi Avanzati per la Formazione Scienter
Italy
Address
Via Val D'aposa 3
40123 Bologna
National Technical University of Athens
Greece
Address
9,Iroon Polytechniou Avenue
15780 Athens