文档介绍：FUZZY CLASSIFICATION WITH A GIS AS AN AID TO
DECISION MAKING
Sasikala ., Petrou M., Kittler J.
Dept. of Electronic and Electrical Engineering
University of Surrey, Guildford, Surrey GU2 5XH, .
Fax:+44 - 1483 34139
Tel:+44-1483-259801
Abstract
In this paper, we discuss various aspects of using fuzzy classification with
a GIS. In particular, we show how fuzzy membership functions to particular
classes can puted posite posed of lots of smal-
ler regions belonging to different classes and how variables taking values in
ranges with different boundary conditions can be handled in a mathematically
rigorous way. We demonstrate our methodology for the problem of assessing
the risk of desertification of burned forest areas in the Mediterranean region.
1 INTRODUCTION
The objective of the present work is to assess the degree of risk of deserti-
fication of burned forest areas using a fuzzy classification technique. It is important to
estimate the risk of desertification in order to take proper measures for its prevention.
Since the parameters involved in the study are fuzzy in nature and have to be classified by
using fuzzy labels like low, medium, high etc., it is felt that it could be more appropriate
to use fuzzy logic. Moreover, the use of remote sensing techniques and GIS along with
fuzzy logic to evaluate the degree of risk would help an expert in a very efficient planning
of resource allocation and decision making.
Work that has already been done on forest fire includes mapping and monitor-
ing of forest fire areas [5], assessment of vegetation change [8] and restoration of burned
areas [2]. Though there have been published work on assessment of areas affected by
forest fire [8], the concept of vagueness has never been considered. Attempts have been
made to include uncertainty in the data [7], but only in terms of probability functions and
not partial membership functions. There have been attempts to use GIS for the classifica-
ti