Object classification by system identification and feature extraction methods applied to estimation of SEM parameters

Literature on estimating Singularity Expansion Method (SEM) parameters to date has been divided into two distinct areas: one which estimates exponentially damped sinusoids of synthetic pulses when the parameters are known a priori, and the other, estimation of “SEM parameters” from laboratory measurements of complex bodies with no a priori knowledge of the “true” SEM parameters. In this paper, we apply an optimal Instrumental Variables (IV) method to the estimation of SEM parameters for two classes of conducting bodies of revolution (BOR), cones and cylinders, for which the theoretical SEM parameters have been computed. The estimated SEM parameters are compared to the theoretical values for each class, and then classification schemes are proposed by which those parameters are used to distinguish the two target classes