Time-Domain Dictionary for Sparse Representation of Radar High Resolution Range Profile

Abstract

The Dictionary is the most important basis of sparse representation which is the key issue of sparse component analysis and compressed sensing. Most of the existing dictionaries are constructed in frequency domain. This paper presented a novel dictionary constructed in time domain for radar signal representation based on geometrical theory of diffraction (GTD) model. Since Radar signal is distributed non-uniformly in time domain, the low-energy part of time responses can be cut-off with insignificant energy loss. As a result, the time-domain dictionary (TD) can be viewed as a sparse matrix, which can save memory and reduce computation complexity greatly, compared with frequency domain dictionaries. Finally, experimental results demonstrate the effectiveness and superiority of the time-domain dictionary.