Model Based Classification using Multi-Ping Data

This paper proposes a method of target classification using three dimensional (3-D) data. The data consists of multiple realizations (pings) of range versus bearing plots, so the three dimensions of the data are range, bearing and time (or pings). The data is assumed to consist of independent non-identically distributed complex Gaussian noise, and a target. The Target (TGT) is of known constant size (extent in range and bearing) and known speed. The TGT power, and heading are unknown. In the derivation of the classifier a normalization step is necessary and we propose an approach to the normalization of multidimensional (m-D) data. This paper contains the derivation of the classifier, a description of the normalizer, a description of the algorithm that follows from the classifier and simulation results