The Derivation of Simple Poles in a Transfer Function from Real Frequency Information. Part 3. Object Classification and Identification.

摘要

This report is the third in a series of three that evaluate a technique (frequency-domain Prony) for obtaining the poles of a transfer function. The main objective was to assess the feasibility of classifying or identifying ship-like targets by using pole sets derived from frequency-domain data. A predictor-correlator procedure for using spectral data and library pole sets for this purpose was developed. Also studied was an iterative method for reducing the effects of noise and a technique based upon magnitude-only spectral data. (Author)