Fast translation invariant classification of HRR range profiles in a zero phase representation

Abstract

The authors describe a new method for aligning high resolution radar (HRR) range profiles for classification purposes. They describe the effects of translational range migration on the phase of the Fourier transformed profiles, and show how the problem of finding an absolute alignment of HRR profiles can be described as a problem in phase estimation. Using this description, they propose a new alignment method, the time-smoothed zero phase representation, which they compare to existing alignment methods in terms of classification performance. Classification is performed using a training set of simulated profiles to classify both simulated and measured test profiles. The proposed method has two main advantages compared to those described in the known literature. Since translation invariance is achieved early on in the classification process, it becomes possible to use more advanced feature extraction methods. Furthermore, the computational cost of classification is considerably lower compared to traditional alignment methods based on cross-correlation.