Syntactic classification of radar targets using polarimetric signatures

A method for classifying radar signatures using a syntactic classifier is presented. Broadband, full-polarization radar scattering data are processed into a time-domain waveform called a transient polarization response (TPR). The TPR is partitioned into a series of subsignatures, each of which is assigned a polarimetrically defined symbol relating directly to substructure shape. The target strings are classified by means of finite-state automata which have been trained, at a specified level of noise, to recognize the strings most likely to be produced by a given radar target. Simulation studies indicate that low-frequency radar backscatter returns of commercial aircraft can be classified reliably using this method