FuRII - An ENVI Toolbox for Image Classification Part 1: Theory Related to Fuzzy Sets, Evidential Reasoning and Fusion

PDF

Authors
  1. Leduc, F.
Corporate Authors
Defence R&D Canada - Valcartier, Valcartier QUE (CAN)
Abstract
In a context of ATD/ATR based on image analysis, there are many sources of imprecision that can lead to uncertain conclusions. In this context it is very important to have a tool that correctly quantifies the uncertainty of detection results given by a set of different sources. Fuzzy sets and evidential reasoning are well suited for this purpose, but at this time there exists no tool dedicated to target detection based on these concepts. For this reason it was decided to develop such a tool called FuRII (Fuzzy Reasoning applied to Image Intelligence). This document addresses image classification. It describes the foundations of fuzzy sets and evidential theories and the fusion mechanisms offered by these two approaches. It is also demonstrated how fuzzy sets and evidential theories can be combined – by modeling imprecise knowledge with fuzzy sets and by fusing multisource information with the evidence theory. Advantages and disadvantages of both theories and fusion mechanisms are explained in detail and illustrated with numerous examples. This document is the first of a series of three. Document II addresses the development and the implementation of FuRII and Document III addresses the validation of the tool. FuRII was developed in the framework of the IB project as a part of the Sensor Data Processing Group (Spectral and Geospatial Exploitation Section) contribution to this project. FuRII is written in the IDL programming language in order to implement the concepts in ENVI

Il y a un résumé en français ici.

Report Number
DRDC-VALCARTIER-TM-2005-276 — Technical Memorandum
Date of publication
01 Jul 2008
Number of Pages
89
DSTKIM No
CA031061
CANDIS No
529751
Format(s):
Hardcopy;CD ROM

Permanent link

Document 1 of 1

Date modified: