Date of Award

5-1994

Degree Type

Thesis

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical Engineering

Supervisor

Simon Haykin

Abstract

A polarimetric radar navigation (PRAN) system makes use of a specially modified marine radar and polarization rotating twist-grid retroreflectors in order to navigate a confined waterway, even in inclement weather or after dark. Despite the polarization diversity offered by such a radar target, depolarization allows significant cross-polar clutter to obscure the reflector return. The objective of the thesis is to successfully demonstrate the enhancement and detection of a cooperative cross-polar target.

A field experiment is designed in Hamilton Bay, and 28 scans of real-time non-coherent HH-pol and HV-pol radar video recorded in a digital format from atop the Canadian Centre for Inland Waters, in Burlington, Ontario. The two reflectors are located at sites in the Dofasco area and the La Salle Park area. A conventional cell-averaging CFAR processor is initially used to give a benchmark against which to compare joint signal processing methods. A dimensionless normalized target-to-clutter ratio (NTCR) is introduced to quantify performance, along with standard sub-images to subjectively show the effect of the processing.

An adaptive cross-polar interference canceller is designed which processes the dual-polarization channels jointly, reducing the nonstationary clutter variance and enhancing the target. An analog implementation of the processor was granted Canadian and U.S. patents.

In another approach, mutual information based unsupervised learning of linear and nonlinear networks is investigated. The RBF network is shown to greatly enhance cross-polar reflector response in the non-Gaussian statistical environment.

Next, a modular solution integrates all three methods to produce superior reflector enhancement in average and peak clutter.

Finally, a novel post-detection processor is demonstrated that successfully uses a priori information about the reflector location along the water-land boundary of the waterway. A fuzzy processor combines primary detection information with the output from a vision-based edge detector to effectively remove false alarms.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS