FUZZY CLASSIFICATION ALGORITHMS APPLIED TO SAR IMAGES OF SHIPS

PDF

Authors
  1. Tuttle, A.
Corporate Authors
Defence Research Establishment Ottawa, Ottawa ONT (CAN)
Abstract
The resolution of Synthetic Aperture Radar (SAR) images has improved to such an extent that classification of SAR imaged targets, such as ships, is possible. The ability to detect and image ships at long ranges, coupled with the abundance of images that may be acquired, has created the need for a system for automatic classification of vessels at long ranges. The Airborne Radar and Navigation Section at the Defence Research Establishment Ottawa has developed several automated systems, using statistical techniques, neural networks and expert systems, to perform this function. The report documents an attempt to implement such a system using fuzzy logic. Fuzzy methods are particularly adapt at modelling systems which involve data sets with ill-defined categories; in this sense SAR images of ships are a natural application for these techniques. The images display wide variability in their appearance as a result of different ship structures and sea-induced rotational motions of the ship. The latter causes variability even for the same ship at the same orientation. An introduction of fuzzy sets and their roles in pattern recognition research is provided. Several fuzzy classifiers are described, and two are tested on some simple data sets. One, the fuzzy K-Nearest Neighbour algorithm failed to produce satisfactory results. TRUNCATED
Keywords
K-Nearest Neighbour Algorithm;Fuzzy Hyperboxes Algorithm
Report Number
DREO-TN-95-11 — Technical Note
Date of publication
01 Aug 1995
Number of Pages
78
DSTKIM No
96-01456
CANDIS No
155040
Format(s):
Document Image stored on Optical Disk;Hardcopy

Permanent link

Document 1 of 1

Date modified: